python
kafka
pgvector
personal projects built with production-grade thinking. everything links to source.
Model-agnostic memory infrastructure for LLM applications. Any AI client writes memory after a session, queries before the next — BM25 + pgvector + RRF hybrid search. Ships with Python and TypeScript SDKs.
Open-source, self-hostable alternative to Jobright.ai. AI job matching, resume tailoring, insider connections, 1-click autofill extension, and an autonomous apply agent.
Distributed event pipeline for high-throughput log ingestion. 40% latency reduction via Redis stream processing.
URL shortener with custom aliases, click analytics, rate limiting, and async job processing via message queue.
Voice-first AI service with a RAG pipeline using FAISS for document-grounded responses via GPT-4.
AI tool that tailors LaTeX resumes to job descriptions. Supports local models via Ollama and cloud via Claude API.
Live Python-to-UML class diagram visualizer for LLD interview prep. Type Python, see the diagram update in real time.
276 curated ML topics across 17 sections — math foundations to frontier research. Progress tracking, difficulty filters, search. No paywalls, no filler.
426 LeetCode problems ordered by learning progression, not arbitrary difficulty labels. Build real intuition.
Privacy-centric transit payment app with contactless payments, RSA encryption, and GDPR/PIPEDA compliance.
i'm a backend engineer with a Master's in Computer Science from McGill University. i build distributed systems and AI-powered tooling — the kind of work that runs at scale and holds up at 2am. i think end-to-end: architecture, deployment, and what breaks when it does.
i also write about backend engineering for 28K+ followers on LinkedIn — not a side hobby. it's made me better at explaining complex systems clearly, which makes me a stronger teammate.
→ suraj.verma@mail.mcgill.caopen to remote backend engineering and DevRel roles in Canada. happy to talk distributed systems, AI tooling, or content.
📅 book a call