AI-powered Indian legal research and document-evaluation platform built on Next.js + FastAPI. Ingests judgments, statutes, circulars, and form templates via OCR pipelines (Docling + PyMuPDF) with role-based access control.
Hello, I'm
Shivam
Pandey
I build
Software Developer at TechOTD Solutions — shipping production AI products across Next.js, FastAPI, LangGraph, and native Android stacks.
Building AI-powered products
at production scale
Software Developer @ TechOTD Solutions
I'm a results-driven software developer with hands-on experience shipping three production AI products from concept to deployment. I specialize in hybrid RAG retrieval, multi-agent orchestration, and cloud-native architectures across Next.js, FastAPI, LangGraph, and native Android stacks.
My work spans Indian and Taiwan legal AI platforms to Android transcription apps — all built on real infrastructure, serving real users. I care deeply about system design and writing clean, maintainable code.
AI Products
Containers
& Frameworks
Experience & Education
Software Developer
TechOTD Solutions Private Limited
- Owned end-to-end delivery of LegalDrishti, Lexmate, and VoxaAI — three production AI products across Next.js, FastAPI, LangGraph, and native Android stacks.
- Designed hybrid RAG retrieval combining Milvus/Pinecone vector search with PostgreSQL full-text and boolean filtering, fused via Reciprocal Rank Fusion with NLP query understanding and legal entity extraction.
- Built LangGraph multi-agent orchestration routing across Claude, GPT, and Llama 3 backends for statute search, case-law retrieval, and document drafting; shipped a clause-by-clause Document Evaluation engine.
- Integrated Deepgram Nova-3 for multilingual real-time speech-to-text with speaker diarization and Gemini for AI summarization, with Firebase Auth, Firestore sync, and Google Play Billing.
- Containerized platforms on AWS EC2 via Docker Compose (9 containers: Nginx, PostgreSQL, Milvus, Ollama) with GitHub Actions CI/CD enabling zero-touch deployments on every push to main.
B.Tech — Computer Science Engineering
NIET, Greater Noida
- Relevant Coursework: Python, OOP (Java), Data Structures & Algorithms, DBMS, Operating Systems, Computer Networks
Technical Skills
Languages
Frameworks & Runtimes
AI / LLM
Databases
DevOps & Cloud
Featured Projects
Taiwan legal AI assistant using LangGraph multi-agent orchestration with specialized graphs for statute search, case-law retrieval, and document generation across Claude 3, GPT, and Llama 3 backends. Includes memory/reflection agents and LangSmith observability.
Android (Java) app integrating Deepgram Nova-3 for multilingual real-time transcription with speaker diarization and Google Gemini for AI summarization with a follow-up Q&A chat interface. Published with full monetization stack.
More projects and open-source work on GitHub
View GitHub ProfileLet's Work Together
Open to new opportunities, collaborations, and interesting problems worth solving.