Lakshay
Systems & AI Architect
IIT Delhi '27 • Civil Engineering • Computer Science Minor
Building robust systems from the ground up—from Unix shells and high-performance Redis clones to Agentic RAG pipelines and edge AI optimization.
The Systems Forge
High-Performance Redis Clone
Low-latency, in-memory KV store built with raw C++. Implements RESP protocol, non-blocking I/O with poll/epoll, and AOF/RDB persistence.
Custom Unix-like Shell
POSIX-compliant shell with process lifecycle management (fork/exec), pipelines, signal handling, and I/O redirection.
Edge AI Optimization Pipeline
Optimized YOLOv8-tiny for edge hardware using TensorRT INT8 quantization, achieving 128 FPS on Jetson-based GPUs.
The AI Lab
Agentic RAG System
Chat across multiple PDFs simultaneously. Chunks, embeds, and indexes documents into FAISS at upload time — then uses conversational memory to handle follow-up questions naturally. Switch between Llama, Qwen, and other Groq-hosted LLMs mid-conversation without losing context.
Book Recommender System
Full-stack discovery engine powered by collaborative filtering on 270k+ books. Precomputes pairwise cosine similarity across a 706×810 user-book pivot matrix to surface titles that resonate with the same audience — no genre tags, just pure behavioural signal.
The Application Layer
ShopSphere (Flash Sale Engine)
Industrial-grade ordering system handling 1,000+ RPS with Redis-based distributed locking and RabbitMQ queues.
ShipyardDeploy
Vercel-style deployment platform that clones GitHub repos, automates build pipelines, and serves per-project preview URLs with real-time logs.
Technical Toolkit
Languages
- C / C++ (C++17)
- Python (PyTorch/DL)
- JavaScript / TypeScript
- SQL
Systems
- Linux (Debian/Bash)
- Docker & Containers
- Redis / PostgreSQL
- Distributed Systems
AI Research
- RAG Architectures
- LLM Benchmarking
- Fine-tuning (LoRA)
- Vector Databases
Design Tools
- Unity Engine
- System Thinking
- Technical Writing
- Rapid Prototyping