Complete zero-to-interview-ready guide for students targeting all 25 dream companies. Built for real placement outcomes, not theory overload.
Demand strong DSA (LeetCode Hard), deep CS fundamentals, system design, and multiple grueling rounds. Expect 4–6 interview rounds. Bar is extremely high.
Strong DSA + domain knowledge. Finance firms add SQL and quant reasoning. BCG needs case interviews. 3–5 rounds.
Good DSA + OOPs + SQL + fundamentals. Communication and accuracy matter. 3–4 rounds. Finance awareness needed.
Aptitude + basic coding + communication + domain awareness. Good fundamentals + projects can clear these. 2–3 rounds.
| Company Type | Examples | Primary Expectation | Key Skills |
|---|---|---|---|
| Product / FAANG+ | Amazon, Alphabet, Microsoft, NVIDIA | Problem-solving excellence | DSA, System Design, CS Depth, Leadership Principles |
| Enterprise Software | SAP, ServiceNow, IBM | Full-stack + enterprise thinking | OOPs, Databases, APIs, Cloud basics, Product sense |
| Finance / Banking | JPMorgan, HSBC, Fidelity, BofA | Accuracy + reasoning + tech | DSA, SQL, OOPs, Quant aptitude, Finance basics |
| Consulting | BCG, EY, Accenture | Communication + structured thinking | Case studies, aptitude, logical reasoning, basic tech |
| Semiconductor / Hardware | Marvell, SanDisk, RTX | Low-level systems + CS fundamentals | C/C++, OS, Computer Architecture, Embedded basics |
| Healthcare / Domain | Eli Lilly, Stryker, Thomson Reuters | Tech + domain curiosity | Coding basics, domain awareness, communication, projects |
| Service / IT | Infosys, Accenture | Communication + trainability | Aptitude, basic coding, OOPS, SQL, attitude |
There are 1.5 million+ engineering graduates every year in India. The 25 companies on this list hire a tiny fraction. Your CGPA alone won't save you. Your college name alone won't save you. The only thing that separates the placed from the unplaced is what you can demonstrate under pressure in 45 minutes. If you can't solve a LeetCode Medium right now without hints — you're not ready. And "I'll start tomorrow" is the most expensive sentence in placement season.
Every concept you need, why it matters, and how deep to go. Filter by priority.
| Concept | What It Is | Why It Matters | Used In | Difficulty | Priority |
|---|---|---|---|---|---|
| A — Arrays | Contiguous memory blocks; index-based access | Foundation of almost every DSA problem; appears in 95% of tests | All companies | Easy–Med | High |
| A — Aptitude | Quant + reasoning + verbal ability | Mandatory for service, finance, consulting companies to get shortlisted | Infosys, Accenture, EY, JPMC, BCG | Easy–Med | High |
| A — API Design | REST/GraphQL interface design principles | System design rounds, backend project discussions | Amazon, SAP, ServiceNow, IBM | Med | High |
| B — Binary Search | O(log n) search on sorted data | Pattern extends to many optimization problems; very common in tests | Amazon, Google, Microsoft, NVIDIA | Med | High |
| B — Backtracking | Recursion + pruning for combinatorial problems | Permutations, N-Queens, Sudoku — appears in product company OAs | Amazon, Alphabet, Microsoft | Med–Hard | High |
| B — Bit Manipulation | Bitwise operations XOR, AND, OR, shifts | SanDisk, Marvell love this; compact tricks in competitive coding | Marvell, SanDisk, NVIDIA | Med | Med |
| C — Complexity Analysis | Big-O time and space analysis | Every coding round expects you to state and justify complexity | All companies | Easy | High |
| C — Computer Networks | OSI model, TCP/IP, HTTP, DNS, load balancing | System design, backend roles, network-related interview questions | Amazon, Microsoft, IBM, SAP | Med | High |
| D — Dynamic Programming | Memoization / tabulation for optimal substructure problems | Separates average from great candidates; Amazon, Google love it | Amazon, Alphabet, Microsoft, JPMC | Hard | High |
| D — DBMS | Database fundamentals — normalization, ACID, transactions | Backend roles, finance companies, every company with data systems | JPMC, BofA, IBM, SAP, Fidelity | Med | High |
| D — Data Structures | Core structures — arrays, linked lists, trees, graphs, heaps | Foundation of all technical interviews | All companies | Easy–Hard | High |
| E — Embedded Systems | Low-level programming for hardware | RTX, Marvell, SanDisk hardware roles; not needed for software roles | RTX, Marvell, SanDisk | Hard | Med |
| F — Fenwick Tree | BIT for prefix sum queries in O(log n) | Competitive coding, rarely in standard placements | Competitive coding oriented roles | Hard | Med |
| G — Graphs | BFS, DFS, shortest paths, topological sort | Very common in product company interviews; models real-world problems | Amazon, Alphabet, Microsoft, NVIDIA | Med–Hard | High |
| G — Greedy | Locally optimal choices for globally optimal solution | Interval scheduling, activity selection — medium difficulty but high frequency | Amazon, Alphabet, Morgan Stanley | Med | High |
| G — Git / GitHub | Version control basics — commit, branch, merge, PR | All practical job roles; resume and project discussions | All companies | Easy | High |
| H — Hashing | HashMap/HashSet for O(1) lookup | Frequency problems, two-sum, duplicate detection — extremely common | All companies | Easy | High |
| H — Heaps | Priority queues, min/max heap operations | Top-K problems, scheduling — medium-high frequency in OAs | Amazon, Microsoft, Alphabet | Med | High |
| H — HR / Behavioral | STAR-method answers to behavioral questions | Every company has this round; can override technical performance | All companies | Easy | High |
| I — Intervals | Merge, insert, find non-overlapping intervals | Calendar/scheduling problems; medium frequency | Amazon, Google, Fidelity | Med | Med |
| J — JVM internals | Java garbage collection, memory model | Rarely asked unless Java backend role; niche | IBM, SAP Java roles | Hard | Low |
| L — Linked Lists | Singly/doubly linked lists, cycle detection, reversal | Classic interview questions; tests pointer manipulation | All companies | Easy–Med | High |
| L — Linux / Unix Basics | Shell commands, processes, file permissions | All backend and system roles; common in IBM, RTX, HP | IBM, RTX, HP, Microsoft | Easy | High |
| M — Matrix / 2D Arrays | Grid problems, BFS/DFS on matrix | Frequent in OAs; island count, shortest path in grid | Amazon, Alphabet, Marvell | Med | Med |
| N — Number Theory | GCD, LCM, prime sieve, modular arithmetic | Appears in aptitude + math-heavy OAs | JPMC, Morgan Stanley, NVIDIA | Med | Med |
| O — OOPs | Encapsulation, Inheritance, Polymorphism, Abstraction | Design questions, code quality; universal expectation | All companies | Easy–Med | High |
| O — Operating Systems | Processes, threads, scheduling, memory management | System-level questions; very common in product + semiconductor roles | NVIDIA, Marvell, Microsoft, IBM, RTX | Med | High |
| P — Prefix Sum | Precomputed cumulative sums for range queries | Subarray sum problems; extremely common trick | All companies with DSA focus | Easy | High |
| P — Probability | Basic probability + Bayes theorem | Finance companies, NVIDIA, quant-related roles | JPMC, Morgan Stanley, NVIDIA | Med | Med |
| Q — Queue / Deque | FIFO data structures, sliding window problems | BFS, sliding window maximum, stream problems | All product companies | Med | Med |
| R — Recursion | Function calling itself with base case + recursive case | Trees, divide & conquer; foundational for DP and backtracking | All companies | Med | High |
| S — SQL | SELECT, JOIN, GROUP BY, subqueries, window functions | Finance, enterprise software, data roles — very high frequency | JPMC, BofA, IBM, SAP, Fidelity, HSBC | Easy–Med | High |
| S — Sliding Window | Fixed/variable window over arrays/strings | Substring problems, max sum subarrays — pattern is very reusable | Amazon, Alphabet, Microsoft | Med | High |
| S — Stacks | LIFO data structure; monotonic stack patterns | Next greater element, expression evaluation — high frequency | All companies | Easy–Med | High |
| S — Sorting | QuickSort, MergeSort, counting sort; custom comparators | Every interview; must know when to use which and implement from scratch | All companies | Easy–Med | High |
| S — Segment Tree | Range query + update in O(log n) | Rare in placements; appears in competitive tracks at NVIDIA, Marvell | NVIDIA, Marvell, competitive roles | Hard | Med |
| S — System Design | HLD/LLD — design scalable systems | Senior-leaning roles at Amazon, Google, Microsoft, ServiceNow | Amazon, Alphabet, Microsoft, ServiceNow | Hard | High |
| T — Trees | Binary trees, traversals, LCA, diameter, paths | Very high frequency across all product companies; deep topic | Amazon, Alphabet, Microsoft, NVIDIA | Med | High |
| T — Tries | Prefix tree for string matching / autocomplete | Google, Amazon — word search, prefix problems | Alphabet, Amazon, ServiceNow | Med–Hard | Med |
| T — Two Pointers | Two indices moving through array — collision or same direction | Sorted array problems, container problems; very high ROI | All companies with DSA | Easy–Med | High |
| T — Topological Sort | Ordering of DAG nodes; Kahn's algorithm or DFS | Task scheduling, build systems, prerequisite problems | Amazon, Alphabet, Microsoft | Med | High |
| U — Union Find | Disjoint set union for connected components | Kruskal's MST, network connectivity problems | Amazon, Alphabet, Marvell | Med | Med |
| V — Version Control | Git branching, commits, rebasing, pull requests | Every job interview — practical software engineering | All companies | Easy | High |
| W — Web Fundamentals | HTTP, REST, JSON, CORS, cookies, auth basics | Full-stack and backend roles; system design foundation | Amazon, SAP, ServiceNow, IBM, HP | Med | High |
| X — XGBoost / ML | Gradient boosting, ML basics | Data/ML roles at Alphabet, NVIDIA, Eli Lilly research | Alphabet, NVIDIA, Eli Lilly | Hard | Low |
| Y — YAML / Config | Configuration files; CI/CD pipelines | DevOps adjacent roles; not primary interview topic | ServiceNow, IBM, HP | Easy | Low |
| Z — Zero-based Indexing Tricks | Off-by-one, modulo patterns, circular arrays | Common bug pattern; tested in implementation-heavy problems | All coding rounds | Easy | Med |
You can watch 200 hours of YouTube DSA content and still fail your first OA. Passive learning creates an illusion of competence. If you watched the solution before struggling for at least 30 minutes, you didn't learn it — you just consumed entertainment. The students who get placed are the ones who sit with a blank editor, get stuck, feel frustrated, and push through. That discomfort IS the learning. Stop collecting bookmarks and start solving problems.
Six structured phases from zero to interview-ready. Follow in order — don't skip phases.
Master your language's syntax. Solve 2-pointer, prefix sum, and sliding window on arrays. Target 25 problems.
Anagrams, palindromes, frequency maps. Learn merge sort and quicksort from scratch.
Reversal, cycle detection, LRU cache. Monotonic stack patterns.
All traversals, LCA, diameter, height. Binary search on answer space.
Kth element problems. Top-K patterns. Activity selection, interval problems.
BFS, DFS, Dijkstra, Bellman-Ford, topological sort. Word search with Trie.
1D DP → 2D DP → Knapsack → String DP → Trees DP. 50+ DP problems minimum.
HLD/LLD, CS fundamentals revision, company-specific mock OAs.
Daily problems + weekly full interviews + behavioral prep + final revision.
Every year, students panic in August when placements begin. They scramble to learn in 2 weeks what should've taken 4 months. The timeline above isn't optional — it's the minimum. If you're reading this in your final year and you haven't solved 100+ problems yet, you're already behind. Not "a little behind" — significantly behind. The gap between you and a prepared candidate isn't intelligence, it's the 500+ hours of deliberate practice they put in while you were "planning to start."
Every topic with patterns, questions, and company relevance. Study in this exact order.
Make locally optimal choice at each step. Prove correctness via exchange argument. Key: greedy works when there's NO benefit to revisiting decisions.
Every unplaced student says the same thing: "I understand the concepts, I just couldn't solve it in time." That's because understanding ≠ ability. You don't "understand" binary search until you can write it bug-free in 3 minutes without Googling. You don't "know" DP until you can identify the state transition in a new problem you've never seen. Interviewers have seen thousands of candidates. They can tell in 5 minutes whether you've actually practiced or just read about it. There is no shortcut. Solve the problems.
Prefix tree for string operations. Each node represents a character. Supports O(L) insert/search where L = word length.
Linear ordering of vertices in DAG such that for every edge u→v, u comes before v. Two approaches: Kahn's algorithm (BFS-based) and DFS-based.
Study these in this order. Focus on the bold topics first.
Companies don't hire generalists who "explored" every technology. They hire people who can go deep under pressure. If your OS knowledge is "I know what a process is," that's a reject. If your SQL is limited to SELECT *, that's a reject. If you can't explain ACID without reading notes, that's a reject. Depth beats breadth. Master the bold topics in each section above before touching anything else. An interviewer would rather hear "I don't know Segment Trees but I can implement Dijkstra from scratch" than "I've heard of everything but can't code any of it."
Mandatory for Tier 3–4 companies. Speed + accuracy = passing the OA cutoff.
| Company Type | Most Important Aptitude Areas | Min Score Needed |
|---|---|---|
| Service (Infosys, Accenture) | Quant (all) + Logical + Verbal + Basic Coding | 75%+ overall |
| Finance (JPMC, BofA, HSBC) | Probability + P&C + DI + Percentage + Puzzles | 85%+ quant |
| Consulting (BCG, EY) | Case reasoning + DI + Verbal + Estimation | 90%+ accuracy |
| Product (Amazon, Microsoft) | Minimal aptitude — coding skills dominate | Pass threshold ~60% |
| Semiconductor (Marvell, SanDisk) | Technical MCQs + basic quant | 70%+ technical MCQ |
You spent months on DSA but can't pass the aptitude cutoff? Game over before it starts. Infosys, Accenture, and most Tier 3–4 companies eliminate the majority through aptitude. And no, you can't "wing it" — these are timed tests where 1 minute per question is the norm. If you haven't practiced under time pressure, you'll freeze, second-guess, and run out of time. The students who clear these aren't math geniuses — they're the ones who did 20 problems a day for 3 weeks. It's a trainable skill, but only if you actually train.
❌ "I built a todo app." → ✅ "I built a task management system that reduces sprint planning time by eliminating context-switching between 3 tools, used by 50 team members."
Every project must have numbers: "Reduced API latency from 800ms to 120ms using Redis caching." / "Achieved 94.2% accuracy on a 10,000-sample dataset." / "Handled 500 concurrent requests in load testing."
Be ready to answer: "Why did you choose FastAPI over Flask?" / "Why PostgreSQL over MongoDB?" / "What would you improve if you had more time?" These show engineering maturity.
GitHub README must have: problem statement, architecture diagram, setup steps, screenshots/demo link, tech stack, future improvements. Recruiters and interviewers check this.
Free options: Vercel (frontend), Render/Railway (backend), HuggingFace Spaces (ML). A live demo beats any description. Put the link in your resume and GitHub profile.
[Action Verb] + [What you built/did] + [How/Technology] + [Result/Impact]
Example: "Engineered an NLP resume screening pipeline using Sentence-BERT and FastAPI, achieving 91% match accuracy and cutting manual screening time by 70% across a 500-resume benchmark."
Every bullet on your resume should answer: "So what?" and "How much?"
Interviewers have seen the same MERN stack todo app, weather dashboard, and e-commerce clone a thousand times. They know the YouTube tutorials by heart. The moment they ask "why did you choose MongoDB over PostgreSQL?" or "how would you add authentication?" and you freeze — they know you copied it. A real project means you made design decisions, hit bugs you had to debug yourself, and can explain every line. If you can't answer "what would you improve?" with genuine insight, it's not your project — it's someone else's code on your resume. That's worse than having no project at all.
Use the PAR Framework: Problem → Approach → Result
"I built [project name] which [solves this specific problem] for [this user]. I used [key tech stack]. The main challenge was [X] and I solved it by [Y], resulting in [metric]."
Pre-prepare 3–4 numbers for each project. Accuracy %, latency, user count, dataset size, throughput, cost saved, time reduced. These stick in interviewers' memory.
Infosys: Aptitude OA → Coding (2 easy problems) → HR. Accenture: Aptitude + Technical MCQs → Coding → Communication + HR. EY: Case + technical basics + HR. Very structured. Communication is weighted highly.
Amazon: OA (2 hard LeetCode) → 4–6 rounds (coding + LP). Alphabet: Phone screen + 4–5 onsite. Microsoft: OA + HR + 2–3 technical + design. NVIDIA: Heavy OS/CN + coding. Marvell: C/C++ + pointers + system programming. SAP: Enterprise case + coding.
Every company on this list receives 10,000+ applications for a few hundred roles. They have the luxury of being picky. Showing up unprepared, not researching the company, or giving generic answers tells the interviewer you don't care enough. "Why do you want to work here?" is not a throwaway question — it's a filter. The students who get offers are the ones who researched the company's tech stack, recent products, and culture. They tailored their STAR stories. They asked smart questions at the end. Preparation isn't just solving LeetCode — it's showing that you want THIS specific job at THIS specific company.
| Week | Daily Topics (Coding) | Daily Topics (Other) | Daily Time | Weekend |
|---|---|---|---|---|
| Week 1 | Arrays (2-ptr, prefix sum, sliding window) — 5 problems/day | Aptitude: % + TSD + T&W (20 Q/day) + SQL SELECT/JOIN basics | 4.5 hrs | Mock aptitude test (timed) |
| Week 2 | Strings + Hashing + Basic Sorting — 5 problems/day | Aptitude: Probability + P&C + Logical (20 Q/day) + OOPs concepts | 4.5 hrs | Full OA simulation (Infosys-style) |
| Week 3 | Linked List + Stack + Queue + Binary Search — 5 problems/day | CS fundamentals (OS + CN basics) + SQL JOINs + Verbal ability | 5 hrs | Mock interview with a peer |
| Week 4 | Trees + BST basics + Recursion — 5 problems/day | STAR behavioral prep (write 10 stories) + Project rehearsal + Resume final | 5 hrs | Full mock interview (Technical + HR) |
You've read this entire sheet. You've bookmarked resources. You've made a Notion page. You've told your friends you're "starting seriously from Monday." None of that counts. The only thing that counts is problems solved, code written, and mock interviews completed. Every day you "plan" without executing, someone else is solving the problem that will appear in your OA. Stop organizing. Stop optimizing your workflow. Open LeetCode right now and solve one problem. Then another. Then another. That's the only plan that works.
| Rank | Topic | Why #1 | Who Needs It | Time to Master |
|---|---|---|---|---|
| 1 | Arrays + Two Pointers + Sliding Window | Appears in 95% of all OAs. Highest ROI. | Everyone | 2 weeks |
| 2 | Hashing + Strings | Second most frequent category. Enables fast solutions to brute-force problems. | Everyone | 2 weeks |
| 3 | OOPs Fundamentals | Universal expectation. Asked in every technical round regardless of role. | Everyone | 1 week |
| 4 | Trees + Binary Search + Recursion | 50%+ of medium-level questions are tree/BS based. | Product + Finance companies | 3–4 weeks |
| 5 | SQL (Basic to Intermediate) | Finance companies have dedicated SQL rounds. Every backend role needs it. | Finance + Enterprise + Backend | 2 weeks |
| 6 | Dynamic Programming | Differentiates average from exceptional candidates at Tier 1 companies. | Amazon, Google, Microsoft, JPMC | 5–6 weeks |
| 7 | Graphs (BFS, DFS, Shortest Path, Topo Sort) | Very high frequency at product companies. Models real-world problems. | Tier 1 + Tier 2 companies | 3 weeks |
| 8 | Aptitude + Reasoning | Primary filter for service + consulting companies. Non-negotiable. | Infosys, Accenture, EY, BCG, Finance | 4 weeks daily practice |
| 9 | System Design (HLD) | Needed for Amazon, Google, Microsoft senior-leaning roles. | Tier 1 companies | 4–6 weeks |
| 10 | Heaps + Greedy + Stack | Medium-high frequency. Enables solving top-K and scheduling problems efficiently. | Tier 1–2 companies | 2 weeks |
| 11 | DBMS + OS + CN Fundamentals | Asked heavily in Tier 1–2. Many behavioral questions tie to OS/DB concepts. | All technical roles | 3–4 weeks |
| 12 | Projects + Resume | Gate-pass for getting shortlisted. Defines interview trajectory. | Everyone — ongoing | Ongoing |
| 13 | Tries + Union Find + Bit Manipulation | Medium frequency. Needed for Marvell, SanDisk, NVIDIA competitive roles. | Semiconductor + Competitive-coding focused | 1–2 weeks |
| 14 | Segment Tree / Fenwick Tree | Rare in standard placements. Only for competitive programming track. | NVIDIA, Marvell advanced roles | Skip unless targeting these specifically |
Placement season is brutal. You will watch people around you celebrate while you refresh your email. You will wonder if you're good enough. That feeling is real, and it's coming for everyone who isn't prepared. But here's what nobody tells you: the students who get placed first aren't always the smartest — they're the ones who started preparing when it was still uncomfortable to study instead of hanging out.
Now close this sheet, open your IDE, and go earn your placement. 🔥