R
Available · Q2 2026
(01) · Portfolio / Rajat Raghav

Agentic AI,shipped solo, atenterprise scale.

I’m Rajat, a product engineer shipping three production AI systems that process 1M+ contracts and cut legal review time by 60%.

Scroll↓
Agentic AI✦Contract intelligence✦LangGraph state machines✦Agentic RAG✦FastAPI at scale✦Distributed systems✦Knowledge Graphs✦Sole-engineer shipping✦
(02) · Selected work

Three systems, one operator.

Jan 2024 – now · Built and owned end-to-end.
01
Enterprise scale

Bulk Analyzer

Agentic AI Analytics System

An AI-driven bulk document analytics platform for an enterprise contract-intelligence client, processing 1M+ contracts through agentic workflows, semantic clause compliance, and risk anomaly detection.

  • LangGraph state machines with cyclic reasoning, fallback nodes, and contextual state hydration
  • Custom load balancer over RabbitMQ + Celery turning blocking ingestion into parallel ML inference
  • MongoDB vector traversal to resolve deeply nested Contract Family hierarchies semantically
  • Interactive Knowledge Graphs via React Flow + Redux surfacing cross-document entity relationships
1M+
Contracts processed
−60%
Legal review time
Days → min
Review cycles
LangGraphLangChainAgentic RAGQdrantFastAPIRabbitMQCeleryMongoDBReact FlowAWS
2024 – now
Agentic AI · Distributed systems
02
Precision · Real-time

Contract Review Engine

Compliance & Risk Analysis

The precision counterpart to the bulk analyzer: a single-contract deep-analysis engine delivering real-time, clause-level legal intelligence via agentic AI workflows.

  • Async FastAPI streaming endpoints for zero-UI-blocking compliance feedback under concurrent load
  • Agentic RAG with Qdrant + Sentence Transformers extracting terms, schedules, and obligations from legal prose
  • Semantic risk scoring engine cross-referencing clauses against corporate playbooks
  • Multi-perspective analysis system evaluating contracts from distinct negotiating stances
−60%
Legal review time
5×
Faster risk flagging
0
UI blocking
FastAPILangChainQdrantSentence TransformersAgentic RAGNLPMongoDB
2024 – now
Product engineering · NLP
03
Agentic · Multilingual

Contract & Proposal Generator

Agentic Document Generation

Industry-grade AI platform assembling legally-compliant contracts and proposals in under 10 seconds, across 1,000+ templates, 25+ languages, and multiple jurisdictions.

  • Agentic generation engine producing compliant contracts from 1,000+ templates in <10 seconds
  • Multilingual NLP pipelines supporting 25+ languages with jurisdiction-aligned clause insertion
  • MongoDB-backed template system with dynamic clause substitution and rule-based compliance
  • Async FastAPI background workers with Pytest-validated error recovery
<10s
Generation time
1,000+
Templates
25+
Languages
FastAPILangChainAgentic AIPrompt EngineeringMongoDBReact.jsPytest
2024 – now
Platform engineering · NLP
Download full résumé →
(03) · About

A product engineer who ships the whole loop.

I spent the last two years building contract intelligence systems for enterprise legal teams. Client meetings on Monday, shipped by Friday.

The work lives at the intersection of language models, information retrieval, and distributed systems. LangGraph state machines, Agentic RAG over Qdrant, RabbitMQ + Celery under FastAPI, all running on AWS behind OpenTelemetry traces.

Shipping production AI alone means every design decision has your name on it. That’s the job I signed up for.

01

Agentic AI systems

LangGraph state machines, cyclic reasoning loops, fallback nodes, and multi-agent orchestration built for unstructured legal and enterprise data.

02

Retrieval & RAG

Qdrant, FAISS, Sentence Transformers, and prompt-engineered LLM calls tuned for clause-level extraction and compliance cross-referencing.

03

FastAPI at scale

Async streaming endpoints, real-time SSE, and tenant-aware middleware that stays calm under concurrent LLM load.

Capabilities
  • i.Agentic AI systemsLangGraph state machines, cyclic reasoning loops, fallback nodes, and multi-agent orchestration built for unstructured legal and enterprise data.
  • ii.Retrieval & RAGQdrant, FAISS, Sentence Transformers, and prompt-engineered LLM calls tuned for clause-level extraction and compliance cross-referencing.
  • iii.FastAPI at scaleAsync streaming endpoints, real-time SSE, and tenant-aware middleware that stays calm under concurrent LLM load.
  • iv.Distributed infraRabbitMQ + Celery with custom load balancing, OpenTelemetry tracing, and AWS-native CI/CD for zero-downtime deploys.
(04) · Principles

Four rules I don’t negotiate.

i.

Systems over scripts

Every shipped thing is a system with a state graph, a failure mode, and a retry policy.

ii.

Boring infra, sharp product

Use the dullest possible infrastructure; save the cleverness for the product surface.

iii.

Evals over vibes

A model is not done because it looks right. It is done when the eval suite agrees.

iv.

Own the whole thing

Sole-engineer shipping taught me that the loop (client call → design → code → deploy → learn) must belong to one person.

“The best software feels inevitable, like the problem could only have been solved this way. Everything else is scaffolding.”
personal memo, 2024
(05) · Experience

Two years, one company, three shipped systems.

Jan 2024 – now
Product Engineer
AI SaaS Startup · Remote · India

Sole engineer on an enterprise engagement. Shipped three production AI systems: Bulk Analyzer, Contract Review Engine, and Contract Generator. Processing 1M+ documents and cutting legal review time by 60%.

LangGraphFastAPIQdrantRabbitMQAWS
(06) · Stack

Tools I trust to run hot.

Deliberately small. Learned deeply. Chosen because they survive incident review at 3 a.m.

Python/TypeScript/FastAPI/Node.js/React.js/Next.js/LangChain/LangGraph/Agentic RAG/Qdrant/FAISS/MongoDB/MySQL/RabbitMQ/Celery/Docker/AWS/OpenTelemetry/
Docker·Jenkins·NGINX·AWS EC2·S3·CloudFront·Elasticsearch·IAM·Cloudflare·Pytest·
AI & LLM
  • LangChain
  • LangGraph
  • Agentic RAG
  • Qdrant / FAISS
  • Sentence Transformers
Backend
  • Python
  • FastAPI
  • Node.js
  • Async Processing
  • Pytest
Distributed
  • RabbitMQ
  • Celery
  • Load Balancing
  • OpenTelemetry
  • Microservices
Cloud & Infra
  • AWS EC2 / S3
  • Amplify · CloudFront
  • Docker · Jenkins
  • NGINX · CI/CD
  • IAM · Cloudflare
(07) · Questions

Anything else you’d like to ask?

A few things I’m asked often. If your question isn’t here, write to me. I answer every legitimate inbound within 48 hours.

(08) · Contact

Got somethingworth building?

Agentic AI delivery, contract intelligence, or platform rescue. If the problem is hard and the team is honest, write to me.

Or call: +91 99717 94844