PORTFOLIO / 2025

Pratik
Sontakke

Agentic AI Engineer with 3+ years of experience in building autonomous multi-agent systems and scalable RAG architectures. Specializing in orchestrating stateful LLM workflows using LangGraph, LangChain, and Python.

Leveraging deep backend engineering principles (Java/Spring Boot) to architect robust, high-performance AI systems. Proven ability to deploy secure, self-correcting AI agents on AWS and Kubernetes for complex tool-calling and real-time traffic.

Download Resume

Open to roles in Agentic AI, Fine-tuning (PEFT), ML, AI Engineering, and collaborations on RAG/Agents projects

India • Remote

RECENT

Agentic AI Engineer - 5C Network

Sept 2025 — Present

FOCUS

LLMsRAGAI AgentsMCPLangChainLangGraphRagasDeepEvalVector DBsFine-tuning (PEFT)FastAPIPythonPostgreSQLSpring BootJavaAWS

Technical Skills

Agentic AI & GenAI

Multi-Agent Systems (LangGraph), LangChain, Model Context Protocol (MCP), RAG Pipelines, LLM Evaluation (Ragas), Tool Calling, OpenAI/Anthropic APIs

Backend & APIs

Python (FastAPI, Pydantic), Java (Spring Boot), Microservices, REST APIs

Databases & Storage

PostgreSQL (pgvector), Redis, MySQL, SQL • Vector DBs (Chroma, Pinecone)

Cloud & DevOps

AWS (ECS, Lambda, Bedrock), Docker, Kubernetes (k8s), Terraform, GitHub Actions, CI/CD

Projects

Enterprise RAG Orchestration Platform

Architected a scalable RAG pipeline that ingests and indexes multi-format data, providing production-grade conversational intelligence for enterprise clients.

Designed a secure Multi-Tenant Architecture on AWS using schema-per-tenant PostgreSQL and RBAC, ensuring strict data isolation within LLM context windows to prevent information leakage.
Developed a high-performance FastAPI orchestration layer utilizing PostgreSQL (pgvector) for vector similarity search, optimizing retrieval latency to enable real-time, context-aware interactions.

FastAPIPostgreSQL/pgvectorAWSRAGMulti-Tenant Architecture

GitHub

Professional Experience

Agentic AI Engineer - 5C Network

Sept 2025 - Present | Bengaluru, India

Architected an autonomous Multi-Agent Supervisor System using LangGraph to audit 5,000+ daily radiology reports, implementing a human-in-the-loop workflow for low-confidence cases.
Integrated Ragas evaluation pipelines to continuously monitor agent "faithfulness" and "answer relevancy," reducing manual QA workload by 30% through automated scoring.
Deployed a context-aware RAG system for internal technical support, optimizing retrieval latency by utilizing Hybrid Search (Keyword + Vector) on proprietary medical data.

AI & Backend Consultant - Consultant

Nov 2024 - Jun 2025 | Remote

Engineered a Text-to-SQL Agent integrated into a legacy Spring Boot microservice, empowering non-technical teams to query complex databases via natural language.
Implemented Model Context Protocol (MCP) patterns to standardize tool-calling interfaces between the LLM and internal APIs, improving system modularity.
Modernized core infrastructure on AWS using Terraform and containerized key Java microservices, establishing a CI/CD pipeline that slashed deployment times to under 10 minutes.

Software Engineer - Guenstiger

Apr 2023 - Oct 2024 | Delhi, India

Optimized high-throughput backend services using Java/Spring Boot, increasing system throughput by 15% to support user growth without additional infrastructure costs.
Developed an early Generative AI pipeline using the OpenAI API to auto-generate hundreds of SEO-optimized product descriptions, resulting in measurable improvements in organic search visibility.

Software Engineer - Edifition

Aug 2021 - Apr 2022 | Remote

Developed and maintained backend modules for client-facing applications, primarily utilizing Java and established standards for service communication.
Improved system stability and reduced development rework by rigorously adhering to API contracts and technical specification standards across assigned projects.