Department of Computer Science & Engineering · PSIT Kanpur
AI / ML · Agent Systems · Blockchain · Mentoring CS students
I teach the engineering behind real AI systems — from clean data structures up to multi-agent workflows. Outside the classroom I build, research, and help students turn course ideas into working software they can show.
National Institute of Technology, Durgapur
CGPA: 9.0Birbhum Institute of Engineering & Technology, West Bengal
CGPA: 9.32West Bengal Board of Secondary Education
86%West Bengal Board of Secondary Education
91.4%Leading engineering at SRM Eco-Tech, Delhi — owning system design and delivery, and driving the technology behind the product from architecture to production.
Current rolePranveer Singh Institute of Technology (PSIT), Kanpur. Taught core CS and applied AI to undergraduate students, advised projects, and built a small research stream around agent systems and ML for real-world problems.
AcademiaRepresented the institute in training and placements — coordinating between companies, students, and the department while completing my M.Tech on campus.
Campus leadershipBuilt the Mission Capable Vehicle Prediction System at Panagarh Military Station — predictive maintenance around military vehicle health, analytics, and field-ready dashboards.
Defence techCryptography and blockchain research (remote) — privacy-preserving verification groundwork that now feeds my ZK-proof and knowledge-graph work.
ResearchBackend systems, enterprise workflows, microservices, and test engineering at Noida — the kind of production work that quietly teaches you what real engineering feels like.
IndustryWhat I teach, what I work on, and how my classroom connects to real software engineering.
From arrays and recursion to graphs, dynamic programming, and complexity analysis. Taught with a practical, interview-aware lens so students leave with working intuition, not just theory.
Classical ML through modern LLMs and agent systems. Students build, evaluate, and ship small AI systems instead of just running notebooks. Strong emphasis on the engineering around the model.
Designing multi-agent workflows with memory, tool use, and evaluation — moving LLMs from demos to dependable systems.
Predictive analytics, computer vision, and grounded retrieval-augmented systems for real-world domains.
Privacy-preserving verification using zero-knowledge proofs over knowledge graphs and on-chain inheritance systems.
Mentoring undergraduates from idea to deployable artifact — the kind of work that builds confidence and a portfolio.
We pick something you actually want to solve, not a textbook clone. The friction makes the learning stick.
A working v1 in days, not weeks. Get to running code first — the architecture conversation happens after.
Define one metric. Measure it weekly. Decisions become data-driven instead of opinion-driven.
Deploy it, write the README, present it. The students who can explain their work clearly are the ones who get hired.
Built from real-world problems. Click any project for full details.
Notes, guides, and curated resources I share with my students — from solving real problems to working with modern AI agents.
The books that have shaped how I teach, think, and build.
A memory-based snapshot of the rounds I cleared, the DSA themes I handled, and the habits that kept saving me.
The pattern that helped me most was simple: explain the brute-force idea first, improve it calmly, test edge cases out loud, and never let panic write the final answer.
TCS rounds felt like a test of clean basics. The trick was not doing magic, just being reliable under pressure.
Wipro felt like one of those rounds where consistency mattered more than trying to look brilliant.
The technical rounds went well — bit manipulation, recursion, complexity tradeoffs. Made it through to HR and didn't make it past. A reminder that how you present yourself matters just as much as how you code.
This one rewarded practical thinking. Clean fundamentals and a calm walkthrough did most of the heavy lifting.
HCLTech was where preparation met opportunity. It was one of the first times I felt interview practice turning into something real.
This experience felt different. It was less about textbook answers and more about whether my thinking could solve something meaningful.
PwC tested whether I could think like an engineer and explain like a consultant. Clean logic mattered, but so did making the solution easy to follow.
The bar here was depth. Every question kept going one level further — not just how it works, but why, and what breaks at scale.
The Prime track raises the bar over the standard TCS rounds — tighter time limits, harder coding sections, and deeper follow-up questions.
Made it through every technical round — DSA, coding, and system questions. The HR round was where it ended. Taught me that technical skill alone is not the whole game.
Cleared the DSA round confidently. The DBMS Administration round went deeper than I was prepared for — it was a valuable lesson in knowing the full stack of what you claim to know.
A few frames from the road that led to teaching — the people, places, and milestones I keep close.
Open to research collaborations, student mentoring, talks, and applied AI projects.