Backend Engineering
APIs that hold up under load — clean architecture, strict types, and databases modeled right the first time.
- ✓ Node.js · Fastify · Express · Django
- ✓ PostgreSQL · MongoDB · pgvector
- ✓ Auth, caching & background jobs
Full-stack developer and agentic AI engineer with a strong backend focus — building scalable web apps, RAG pipelines and LLM agents. GSSoC ’24 maintainer · 500+ DSA problems solved · RAG pipelines, LLM agents & cloud infra in production.
RAPTOR indexing · RRF fusion · live on GitHub
Explore →CURRENTLY INTO
Agentic pipelines, RAPTOR retrieval & distributed inference. Building in public on GitHub.
“Led 50+ contributors as a GSSoC ’24 project admin — 50+ PRs merged, zero-downtime deploys.”
OPEN SOURCE · 70 REPOS
I'm Paramveer — a full-stack developer and agentic AI engineer with a strong backend focus. I build scalable web apps, production RAG pipelines and LLM agents with the unglamorous parts done right: provenance, evals, observability and infra as code.
As a GSSoC '24 project admin I led 50+ contributors and kept deploys green at zero downtime. These days I'm building AI systems in public — open source, documented trade-offs and all.
Senior-level ownership from first sketch to production deploy — and after.
APIs that hold up under load — clean architecture, strict types, and databases modeled right the first time.
From naive chatbot to grounded system: hierarchical indexing, fused retrieval, rerankers and provenance you can audit.
Reproducible infra and pipelines — provisioned as code, observed with dashboards, deployed with zero downtime.
Next.js products end to end — from schema to deploy — fast, type-safe and easy for teams to extend.
Production-grade RAG service with RAPTOR hierarchical indexing (UMAP + HDBSCAN), vector + BM25 search fused via RRF, Jina reranking and Gemini 2.0 Flash answers with citations. Containerized microservices with Prometheus + Grafana observability.
Clinical document agent turning scanned patient records into structured discharge summaries. OCR fallback, deterministic reconciliation, medication-conflict detection and full provenance — every fact traces to source doc, page and text.
Multi-EC2 model inference pipeline built for Alchemyst AI — private-subnet networking, RPC worker coordination, Terraform provisioning and honest latency trade-off docs under t3.micro constraints.
Lightweight AI assistant with streaming UX, rich Markdown (tables, math, code), syntax highlighting with one-click copy and local chat history. Live in production.
Resume analyzer scoring against 8+ ATS criteria — keyword density, formatting, section completeness — with 5+ actionable fixes per resume and a 0–100 recruiter-simulation score. Live in production.
Type-safe note-taking app (98% TypeScript) — rich-text editing, organization, search and Gemini-assisted workflows on Supabase Postgres. Clean, modular, tested.
70+ repos and counting — explore the full archive on GitHub →
Designing and shipping AI systems in the open: a RAPTOR-indexed RAG API with full observability, a clinical document agent with provenance tracking, and distributed inference infra on AWS.
Led an open-source web platform with 50+ contributors. Reviewed and merged 50+ PRs, ran CI/CD with GitHub Actions at zero downtime, and mentored contributors in React, Node.js and Git workflows.
Core CS fundamentals plus 500+ DSA problems across platforms — LeetCode peak 1459, CodeChef 1300. The problem-solving base behind everything above.
Across LeetCode, CodeChef and competitive programming platforms.
With a 1300 CodeChef rating alongside — consistency, not luck.
As GSSoC ’24 project admin — 50+ PRs reviewed, merged and shipped.
Computer Science at Bengal College of Engineering and Technology.
Yes — open to internships, freelance projects and full-time backend / AI engineering roles. I reply within 24 hours.
TypeScript + Python. Next.js and Fastify/Express on the backend, PostgreSQL/pgvector for data, Docker + AWS + Terraform for infra — and RAG/agent pipelines with Gemini and Jina.
Yes — I was a GSSoC ’24 project admin leading 50+ contributors, and most of my AI systems work is public on GitHub.
Absolutely — hit the “Download résumé” button or email me and I’ll share it right over.
Backend builds, AI systems, or just to say hi — I reply within 24h.