AI/ML Engineer building reliable agentic AI systems, shipped as production web APIs

I turn messy, conflicting data into reliable agentic AI systems: refined with NLP, RAG, H3 geospatial indexing and confidence scoring, orchestrated with LangGraph and MCP, and shipped as FastAPI web APIs on GCP and Azure.

Get in touch View case studies

About me

Practical AI, built to survive production

I’m an AI/ML Engineer based in Indore, currently at EngineeringBabu. I design agentic AI systems, LLM applications and the data and API layers that keep them dependable.

I like problems where the answer has to be defensible: research engines that cite evidence, track contradictions and score confidence, and multi-agent workflows with real budgets and guardrails. I work across the current agent stack: MCP and A2A for connectivity, memory and vector search, evals and tracing, and guardrails against prompt injection.

I started as a backend engineer (SDE-II at Systango, then Valere) and studied at Medi-Caps University, Indore (2019–2023, CGPA 8.8). Outside work I enjoy travelling and riding.

Toolbox

PythonFastAPIDjangoFlaskSQLAgentic AILLMsRAGLangChainLangGraphMCPOpenAI Agent SDKLLM fine-tuningPrompt engineeringNLPH3 / geospatialApache IcebergGCPAzureDockerGitPytest

Agent stack & LLMOps

Context engineeringA2A protocolCrewAILlamaIndexReAct / Plan-and-Execute / ReflexionAgent memory (Mem0)pgvectorQdrantChromaLLM evals (DeepEval, Ragas, promptfoo)LangSmithLangfuseGuardrails & prompt-injection defenceMLOpsPyTorch

Selected work

Case studies

Open one to see what I actually built.

AI/ML Engineer · PropTech · Agentic AI

D-Plan: land & property intelligence, backed by evidence

An AI platform that pulls government land records, listings, news, market signals and geospatial data into evidence-backed property insights.

What I built

Agentic research engine that interprets intent, orchestrates tools and searches, retrieves live data, spots information gaps and iterates to a structured research brief.
Land & geospatial layer using MP Bhulekh, H3, location resolution, parcel/Khasra data, roads, infrastructure and terrain, with reusable adapter/protocol abstractions.
Evidence graphs, source and contradiction tracking, confidence scoring, ingestion/EDA pipelines and Iceberg storage.
Agent reliability controls: budgets, retries, timeouts and token limits.

AI Engineer · Procurement · Multi-agent

Nomia: AI-driven procurement, end to end

An agentic B2B procurement platform covering sourcing, vendor management and RFQs, connecting buyers with verified suppliers.

What I built

Multi-agent procurement workflows built with the OpenAI Agent SDK.
Workflows that adapt to a company’s processes, purchase history and compliance rules.
Optimised the workflow to handle a high volume of requests efficiently.

Backend & AI Engineer · Real-time chat · LLMs

Tfonia: real-time chat with AI replies for hosts

A chat and video app where hosts and users connect, and hosts can reply using an AI assistant.

What I built

Fine-tuned and integrated an LLM that replies based on conversation history.

Experience

Where I’ve worked

Jun 2026 – present

AI/ML Engineer

EngineeringBabu

Built the AI/ML team and operations from scratch, and handled multiple projects at the same time as a tech lead.

Mar 2026 – May 2026

Software Engineer

Valere

Built product-level thinking while working in a large team and mentoring juniors.

Jan 2023 – Feb 2026

SDE-II

Systango

Learned how products are developed end to end — how SDLC works and the role each component plays in a product.

Process

How I work, from idea to shipped

A simple loop I use for AI projects.

Presenting a RAG architecture diagram to a product team

Step 1

Seed

Understand & scope

Start from the real problem and the data that exists.

Work with stakeholders on business flows
Define architecture and roadmap
Plan POCs in phases

Designing, building, and launching together as a team

Step 2

Grow

Build & evaluate

Ship small, test hard, and make agents measurable.

Modular agent architectures and tool/adapter abstractions
Custom test harnesses and Pytest for LLM output
Clean REST APIs and data pipelines

Hardening production systems and handing over to the team

Step 3

Flourish

Harden & hand over

Make it dependable once real traffic arrives.

Budgets, retries, timeouts and token controls
Confidence scoring and source tracking
Mentoring and documentation for the team

Have an AI problem worth solving?
Get in touch

Values

What roots everything I do

Evidence first

Outputs should be traceable to sources, with confidence you can inspect.

Reliable by design

Agents need budgets, retries and timeouts like any production service.

Mentor as you go

I help interns and freshers grow and have helped hire backend engineers.

Clear communication

Technical decisions land better when everyone understands them.

Get in touch

Let’s build something intelligent

Open to AI/ML and agentic AI roles and projects.

Email
animesh.u.mishra20@gmail.com

Phone
+91 7999154939

Portfolio

Loading portfolio…

Practical AI, built to survive production

I’m an AI/ML Engineer based in Indore, currently at EngineeringBabu. I design agentic AI systems, LLM applications and the data and API layers that keep them dependable.

I started as a backend engineer (SDE-II at Systango, then Valere) and studied at Medi-Caps University, Indore (2019–2023, CGPA 8.8). Outside work I enjoy travelling and riding.

Case studies

Open one to see what I actually built.

AI/ML Engineer · PropTech · Agentic AI

D-Plan: land & property intelligence, backed by evidence

An AI platform that pulls government land records, listings, news, market signals and geospatial data into evidence-backed property insights.

What I built

Agentic research engine that interprets intent, orchestrates tools and searches, retrieves live data, spots information gaps and iterates to a structured research brief.
Land & geospatial layer using MP Bhulekh, H3, location resolution, parcel/Khasra data, roads, infrastructure and terrain, with reusable adapter/protocol abstractions.
Evidence graphs, source and contradiction tracking, confidence scoring, ingestion/EDA pipelines and Iceberg storage.
Agent reliability controls: budgets, retries, timeouts and token limits.

AI Engineer · Procurement · Multi-agent

Nomia: AI-driven procurement, end to end

An agentic B2B procurement platform covering sourcing, vendor management and RFQs, connecting buyers with verified suppliers.

What I built

Multi-agent procurement workflows built with the OpenAI Agent SDK.
Workflows that adapt to a company’s processes, purchase history and compliance rules.
Optimised the workflow to handle a high volume of requests efficiently.

Backend & AI Engineer · Real-time chat · LLMs

Tfonia: real-time chat with AI replies for hosts

A chat and video app where hosts and users connect, and hosts can reply using an AI assistant.

What I built

Fine-tuned and integrated an LLM that replies based on conversation history.

Where I’ve worked

Jun 2026 – present

AI/ML Engineer

EngineeringBabu

Built the AI/ML team and operations from scratch, and handled multiple projects at the same time as a tech lead.

Mar 2026 – May 2026

Software Engineer

Valere

Built product-level thinking while working in a large team and mentoring juniors.

Jan 2023 – Feb 2026

SDE-II

Systango

Learned how products are developed end to end — how SDLC works and the role each component plays in a product.

How I work, from idea to shipped

A simple loop I use for AI projects.

Step 1

Seed

Understand & scope

Start from the real problem and the data that exists.

Work with stakeholders on business flows
Define architecture and roadmap
Plan POCs in phases

Step 2

Grow

Build & evaluate

Ship small, test hard, and make agents measurable.

Modular agent architectures and tool/adapter abstractions
Custom test harnesses and Pytest for LLM output
Clean REST APIs and data pipelines

Step 3

Flourish

Harden & hand over

Make it dependable once real traffic arrives.

Budgets, retries, timeouts and token controls
Confidence scoring and source tracking
Mentoring and documentation for the team

Have an AI problem worth solving?
Get in touch

What roots everything I do

Evidence first

Outputs should be traceable to sources, with confidence you can inspect.

Reliable by design

Agents need budgets, retries and timeouts like any production service.

Mentor as you go

I help interns and freshers grow and have helped hire backend engineers.

Clear communication

Technical decisions land better when everyone understands them.

AI/ML Engineer building reliable agentic AI systems, shipped as production web APIs

Practical AI, built to survive production

Toolbox

Agent stack & LLMOps

Case studies

D-Plan: land & property intelligence, backed by evidence

Nomia: AI-driven procurement, end to end

Tfonia: real-time chat with AI replies for hosts

Where I’ve worked

AI/ML Engineer

Software Engineer

SDE-II

How I work, from idea to shipped

Seed

Understand & scope

Grow

Build & evaluate

Flourish

Harden & hand over

What roots everything I do

Evidence first

Reliable by design

Mentor as you go

Clear communication

Let’s build something intelligent

Portfolio

AI/ML Engineer building reliable agentic AI systems, shipped as production web APIs

Practical AI, built to survive production

Toolbox

Agent stack & LLMOps

Case studies

D-Plan: land & property intelligence, backed by evidence

Nomia: AI-driven procurement, end to end

Tfonia: real-time chat with AI replies for hosts

Where I’ve worked

AI/ML Engineer

Software Engineer

SDE-II

How I work, from idea to shipped

Seed

Understand & scope

Grow

Build & evaluate

Flourish

Harden & hand over

What roots everything I do

Evidence first

Reliable by design

Mentor as you go

Clear communication

Let’s build something intelligent

Sampti: product intelligence POCs