Operational history
Background
Full Stack & AI Engineer: biologically-inspired computer vision (NeuroSight - 100% directional accuracy, sub-100MB runtime memory), production RAG systems (98.5% retrieval precision), and multi-agent orchestration. Shipped 3 PyPI packages (TunePrompt, Genesis Forge, NeuroSight); merged critical fixes to OpenClaw (330k stars) and Anthropic MCP servers.
Experience
Apr 2026 – Present
Mohali, India · On-site
Deployment / experience node
Full Stack & AI Engineer
Girl Power Talk
- ▶Built Expert Witness Tool Legal AI SaaS (React, Node.js, Express): hybrid RAG pipeline (HyDE + BM25 + Pinecone + RRF + Cohere rerank-v3.5) with parent-child chunking achieving 98.5% retrieval precision and 10x faster response vs. prior n8n workflow.
- ▶Reduced per-case AI processing cost by 125x ($2.00 → $0.016) via async BullMQ + Redis job queuing, SHA-256 file deduplication, and multi-layer caching, eliminating redundant re-embedding on unchanged documents.
Jan 2025 – Mar 2026
Remote
Deployment / experience node
Full Stack & AI Engineer
HomeInTown
- ▶Engineered a lead filtration engine and multi-tenant auth system (MPIN, JWT, MSG91) sustaining 1,000+ concurrent events with 99.9% session persistence and zero cross-tenant data leakage.
- ▶Designed and integrated AI Voice API webhook listeners via Socket.io, maintaining real-time state sync across 40+ concurrent agent dashboards with sub-100ms event propagation.
- ▶Architected a Redis split-worker queuing layer and WhatsApp webhook pipeline (360Dialog) for the AI calling CRM; system is now spinning off as an independent SaaS (VaniOps) on Telnyx + Cloud Run.
Jan 2024 – Dec 2024
Noida, India · On-site
Deployment / experience node
Full Stack Java Developer
QSpiders Development Center
- ▶Optimized Spring Boot RESTful services via PostgreSQL indexing and JPA query tuning, cutting API response times by 30% across 5 core endpoints.
- ▶Reduced team development overhead by 40% by introducing generic repository patterns and automated DTO mapping (Lombok + MapStruct) across a 12-module monorepo.
Technical Stack
Ai LlmCapability module
LangChainLangGraphRAG PipelinesSemantic ChunkingPrompt EvaluationLLM Cost OptimizationAgentic SystemsMulti-Agent OrchestrationAnthropic MCPFine-tuning (LoRA/QLoRA)
Embeddings VectorCapability module
OpenAI EmbeddingsGemini EmbeddingsSentence TransformersPineconepgvector
Ai FrameworksCapability module
OpenAI APIAnthropic APIGemini APIGroqHuggingFace Transformers
LanguagesCapability module
TypeScriptPythonJavaJavaScript
BackendCapability module
Node.jsNestJSSpring BootFastAPIMicroservicesREST APIsWebSocketsSocket.io
FrontendCapability module
Next.js 15React 19Tailwind CSS 4Framer MotionRedux ToolkitZustand
Cloud DevopsCapability module
AWS LambdaAWS S3AWS EC2Google Cloud RunDockerGitHub Actions CI/CDSupabase
DatabasesCapability module
PostgreSQLMongoDBRedis (Upstash)MySQL
Testing EvalCapability module
PostmanLLM output evaluationROUGESemantic similarity scoring
Education
Indira Gandhi National Open University (IGNOU)
Master of Computer Applications (MCA)
Jan 2024 – Dec 2025
Maulana Mazharul Haque Arabic and Persian University
Bachelor of Computer Applications (BCA)
2019 – Jan 2023
Certifications
Model Context Protocol (MCP) Specialist
Anthropic
AWS Solutions Architect
Amazon Web Services