Hi, I'm Vedavrath 馃憢
I build
production AI
that ships.
I build production AI systems across LLMs, RAG, Knowledge Graphs and autonomous multi-agent platforms. Creator of AgenticML, an autonomous multi-agent AutoML framework on PyPI.
Vedavrath Pathi
AI Engineer
about
Turning models into products.
AI Engineer with 3+ years of experience shipping production AI systems built on Large Language Models, Retrieval-Augmented Generation, Knowledge Graphs and AWS cloud services. I design enterprise document-intelligence platforms, agentic AI systems and cloud-native ML solutions. I'm the creator of AgenticML, an autonomous multi-agent AutoML framework published on PyPI.
3+
Years building AI in production
100%
Doc classification accuracy (IDP)
80%
Less physician documentation effort
PyPI
Published framework: swayamml
- Pace University2024 - 2026M.S. in Data Science 路 New York, USA
- Indian Institute of Information Technology, Kottayam2019 - 2023B.Tech in Computer Science & Engineering 路 Kerala, India
- AWS Certified Machine Learning - Specialty
- Microsoft Azure AI Engineer Associate (AI-102)
- AWS Certified AI Practitioner
selected work
Things I've built.
From a multi-agent AutoML framework to enterprise Graph RAG and serverless document intelligence. Here's a sample of systems I've shipped.
AgenticML
FlagshipAutonomous multi-agent AutoML framework
An end-to-end AutoML platform where specialized AI agents collaborate to take a raw dataset all the way to a trained, evaluated model: profiling, preprocessing, feature engineering, training, evaluation and reporting. Published on PyPI as swayamml for one-command pipelines.
- Specialized agents for each ML stage, orchestrated as a graph
- One-command execution from raw data to model + report
- Supports both regression and classification workflows
AgentForge
FlagshipBuild & deploy agents from natural language
A platform that turns a plain-language spec into a working multi-agent system. It builds the agent graph, compiles LangGraph runtimes, wires up MCP tools, and publishes the result straight to GitHub. A full builder, compiler and runtime pipeline for agentic apps.
- Builder graph that designs agents from a natural-language brief
- Compiles to runnable LangGraph agents with MCP tool loading
- One-click publish to GitHub with generated project scaffolding
Agent Wallet
FlagshipSpend & policy layer for autonomous agents
A hosted wallet and spend-control layer that lets autonomous AI agents transact safely, with budgets, policy guardrails and auditable activity so agents can pay for tools and services without going off the rails.
- Budget and policy controls scoped per agent
- Auditable transaction history for every agent action
- Designed as a hosted, multi-tenant SaaS
Enterprise Graph RAG Platform
Knowledge graphs + vectors for grounded answers
A Graph RAG platform combining Neo4j knowledge graphs, vector embeddings and LLMs. It extracts entities and relationships from documents and systems, and translates natural language into Cypher for conversational access to enterprise knowledge.
- Natural-language-to-Cypher translation over a live graph
- Entity & relationship extraction from docs, DBs and PM tools
- Reduced hallucinations by fusing graph traversal with semantic retrieval
AWS Intelligent Document Processing
Serverless document intelligence at scale
A serverless IDP platform on AWS Bedrock, Textract, Rekognition, Comprehend, Lambda and Step Functions that automates document classification, entity extraction and PII redaction for enterprise financial workflows.
- 100% document classification accuracy in client evaluation
- 97% sensitive-information redaction accuracy
- Event-driven pipelines for scalable ingestion
Insurance Knowledge Assistant
RAG retrieval + email automation agent
An enterprise RAG platform for insurance policy retrieval, comparison and recommendation across many documents, paired with an AI email agent that drafts customer communications, claim responses and policy correspondence.
- Hybrid retrieval: vector search + metadata filtering
- Context-aware policy recommendations across documents
- Automated claim & customer correspondence drafting
experience
Where I've worked.
AI Engineer 路 GramX
Jun 2024 - Dec 2025 路 India- Built enterprise RAG applications on AWS Bedrock, OpenSearch and vector DBs for natural-language access to organizational knowledge.
- Designed Graph RAG architectures with Neo4j knowledge graphs connecting docs, tickets and business entities for contextual reasoning.
- Developed multi-agent workflows with LangGraph & LangChain for autonomous task execution and workflow automation.
- Built cloud-native AI services on Bedrock, Lambda, SageMaker and S3 supporting scalable enterprise workloads.
Machine Learning Engineer 路 Minfy Technologies
Sep 2023 - Jun 2024 路 Hyderabad, India- Built an Intelligent Document Processing platform with Textract, Rekognition, Bedrock and Comprehend, reaching 100% classification and 97% redaction accuracy.
- Developed an AI medical-documentation platform generating SOAP notes & summaries at 94% transcription accuracy, cutting physician effort by 80%.
- Implemented hybrid RAG conversational AI, improving retrieval relevance and recommendation accuracy by 20%.
- Built SageMaker training & inference pipelines for fine-tuning and deploying transformer models.
Data Engineering Intern 路 Royal DSM
Jul 2022 - Sep 2022 路 Hyderabad, India- Built data pipelines integrating SAP datasets with AWS analytics infrastructure for enterprise reporting and ML.
- Performed large-scale data transformation and validation for downstream analytics.
- Used AWS CDK to automate cloud infrastructure provisioning.
toolkit
Skills & stack.
The technologies I reach for to take AI systems from prototype to production.
Generative AI
Machine Learning
Cloud & MLOps
Data & Databases
Languages & APIs
let's build
Let's ship something intelligent.
I'm open to AI Engineering roles and collaborations. Whether it's RAG, agents, or AutoML, let's talk.