// 04 ARCHITECTURAL CHRONICLE & EVIDENCE•LIVING DOSSIER
AUTONOMOUS WORKSTATION • HARYANA, IN
Craft, Systems & Evidence.
Engineering software from the silicon boundary up: high-throughput Python microservices, pgvector semantic search pipelines, resilient distributed scrapers, and sovereign infrastructure that answers directly to users.
Sahil
Generative AI & RAG Systems EngineerAI/RAG • pgvector 18ms • LOAD 0.24
PROFESSIONAL SUMMARY// SPEC 01/06: AI & RAG
Product-focused AI Engineer specializing in end-to-end Retrieval-Augmented Generation (RAG) pipelines, pgvector similarity search, and autonomous multi-agent tool execution with sub-30ms semantic resolution. Proven track record deploying containerized microservices, prompt engineering benchmarks, and automated evaluation workflows.
PROFESSIONAL EXPERIENCE// 02
Generative AI & Backend EngineerJAN 2024 — PRESENT
Fusionpact TechnologiesNoida, UP, India (Remote / Hybrid)
- RAG Architectures & Multi-Agent Intelligence: Designed and deployed end-to-end RAG pipelines leveraging LLMs (OpenAI, Gemini) and Agno to convert complex unstructured enterprise documents into structured JSON schemas.
- Semantic Caching & Vector Search: Implemented vector embeddings and similarity search using PostgreSQL with pgvector, slashing API latency by 30% (sub-30ms pgvector resolution) and lowering recurring inference costs.
- Automated Ingestion & Microservices: Architected resilient Playwright scraping engines to ingest bids from government portals with automated CAPTCHA handling, deployed on high-concurrency FastAPI backends.
- Prompt Engineering & Model Validation: Formulated advanced few-shot and chain-of-thought prompt pipelines and evaluation benchmarks to prevent hallucinations, ensuring strict output alignment and schema fidelity.
FEATURED PROJECTS & TELEMETRY// ROTATING 2 OF 6
CONTINUOUS UPDATE FEED: ACTIVE DISPATCH
agentic-api • git:mcp-streamAsynchronous tool execution pipelines with pgvector semantic cache.
Agentic AI & RAG Framework (agentic-api)GITHUB REPO ↗
FastAPI, Python, pgvector, Docker, Agno, LLMs
- Architected a containerized multi-agent backend providing asynchronous RAG analysis for multi-format enterprise documents (PDF, DOCX, XLSX).
- Implemented semantic caching and vector indexing using PostgreSQL with pgvector, optimizing response times and slashing compute overhead.
- Integrated session-aware context management and asynchronous streaming responses, achieving sub-30ms latency for semantic document queries.
Production ARM64 Linux InfrastructureLIVE TELEMETRY ↗
Ampere A1 aarch64, Root Podman Quadlets, Caddy TLS, Systemd, WireGuard
- Supervised 16 systemd container Quadlets with automated Let's Encrypt wildcard TLS via host-networked Caddy reverse proxy.
- Engineered zero-trust WireGuard mesh with split-DNS (Blocky) and automated daily offsite SQLite/Postgres mirrors to Arch Linux.
- Maintained 99.99% operational uptime with automated daily backup timers, journald volume caps, and Beszel monitoring hubs.
TECHNICAL COMPETENCIES// 04
| Languages | Python, TypeScript, JavaScript, SQL, Bash, C++, Dart |
| AI & RAG | RAG Pipelines, pgvector, Agno, LLMs (Gemini, OpenAI), Hugging Face, Transformers.js |
| Backend & Web | FastAPI, Next.js, Node.js, React, Redis, REST APIs, Microservices, Playwright |
| DevOps & Infra | PostgreSQL, Docker, Podman Quadlets, Caddy TLS, Linux (Arch/Ubuntu), WireGuard, CI/CD |
| Interpersonal & EI | High Emotional Intelligence, Cross-Functional Alignment, Blameless Post-Mortems, Mentorship |
ACADEMIC BACKGROUND// 05
Master of Computer Applications (MCA) • 8.67 CGPA
CRSU • Jind, Haryana2023 — 2025
B.Sc (Hons) Information Technology • 7.9 CGPA
Kurukshetra University • Haryana2020 — 2023