AI ArchitectPlano, TX, United States

AI systems,built to hold up.

I'm Anudeep Sri Bathina, an AI Architect with 11 years of experience turning GenAI, RAG, and data platforms into reliable production systems, with explicit boundaries, rigorous evaluation, and measurable impact.

Proof, not promises.

A selection of measurable outcomes from production GenAI, machine learning, and data systems.

01
15K+

people served

Enterprise GenAI platform at AT&T

02
4h → <90s

resolution time

Policy queries accelerated through RAG and automation

03
97.2%

extraction accuracy

Across 50K+ document pages per month

04
2.4s p95

RAG latency

At $0.42 per 1K requests on Azure

05
$2.3M

retention savings

Delivered through customer churn modeling

06
8

engineers led

Moving ML deployments from quarterly to weekly

From data foundations to AI architecture.

Eleven years across engineering, technical leadership, graduate research, and enterprise AI. One continuous practice of making complex systems useful.

Research rigor matters. So do deadlines, operating costs, and the people maintaining the system after launch. My work sits at that intersection.

Explore the full journey

AI Architect

Current

AT&T

Architected enterprise GenAI platform serving 15K+ internal users, reducing policy-query resolution time from 4 hours to under 90 seconds. Built Python-based document parsers processing 50K+ pages/month with 97.2% extraction accuracy. Deployed RAG pipeline via FastAPI + Azure AI Search achieving p95 latency of 2.4s at $0.42/1K requests.

2024 to Present

United States

Technical Lead

Capgemini

Led ML engineering team of 8 engineers to deliver 12+ production data pipelines, reducing model training time by 65% through PySpark optimization. Implemented MLOps framework that increased model deployment frequency from quarterly to weekly. Achieved 94% model accuracy in customer churn prediction, saving $2.3M in retention costs.

2019 to 2021

India

Big Data & ML Engineer

GainInsights Solutions

Engineered scalable PySpark pipelines processing 10M+ records/day, improving ETL performance by 78% and reducing job failures by 92%. Developed real-time dashboard serving 500+ concurrent users with sub-second query response. Optimized database queries achieving 3.2x faster report generation for 25+ enterprise clients.

2019 to 2019

India

Data Science Engineer

Cognizant

Led cloud data migration for 6TB of enterprise data across 3 AWS regions, completing 3 months ahead of schedule with 99.8% data integrity. Implemented predictive analytics models that increased sales forecasting accuracy by 43% and reduced inventory costs by $1.1M annually. Automated 85% of manual data migration tasks, saving 40+ hours/week of engineering time.

2015 to 2019

India

Depth where production gets difficult.

The tools matter. The architectural judgment behind retrieval, evaluation, safety, infrastructure, and data flow matters more.

Agentic AI & Orchestration

Multi-agent systems, reasoning engines & tool-use

LangGraph / CrewAIMCP (Model Context Protocol)OpenAI o3 / o4-miniClaude 4 / Gemini 2.5

RAG & Knowledge Architecture

Enterprise retrieval, grounding & search systems

Advanced RAG PipelinesVector DatabasesGraphRAG & Knowledge GraphsEmbeddings & Reranking

AI Safety & Evaluation

Guardrails, observability & production trust

LangSmith / LangfuseGuardrails & GradingPresidio (PII/PHI)Red-teaming & Evals

AI-Native Development

Agentic coding & AI-powered engineering

Claude Code / Codex CLICursor / WindsurfGitHub CopilotA2A (Agent-to-Agent)

Cloud & Production Systems

Enterprise infrastructure built for AI workloads

Python & FastAPIAzure AI / AWS BedrockDocker & KubernetesCI/CD & MLOps

Data Engineering

Scalable pipelines, warehousing & real-time processing

PySpark / DatabricksSQL & dbtAirflow / PrefectKafka & Streaming

Current questions

What I'm exploring now

Agentic AI orchestration with LangGraph + MCP
Multi-modal RAG with vision models
AI safety evaluation frameworks
Building ClinIQ v2 with department-scoped retrieval

Selected systems & open-source work.

Reference implementations that make the architecture, trade-offs, and operating boundaries visible, not just the demo.

ClinIQ

FeaturedOpen source
Healthcare RAG architecture with scoped retrieval and conservative safety controls
Healthcare RAG · citations · PII safety

Healthcare RAG reference implementation with department-scoped retrieval, citations, role checks, and PHI masking.

LangGraphLangChainFastAPIOpenAIChromaDBPresidio

EvidenceIQ

Open source
Privacy-aware media intake, search, and review
local-first · sensitive media · audit trails

Local-first sensitive-media workspace with role-based access, PII-aware metadata handling, local AI tagging, semantic search, and audit trails.

FastAPISQLiteOllamaChromaDBRBACSHA-256

InferIQ

Open source
GPU inference benchmarking, routing, and observability
latency · throughput · GPU memory · cost per token

Reference implementation for benchmarking and serving LLM inference across vLLM, NVIDIA NIM, and NVIDIA NeMo, with optional CUDA profiling, a FastAPI gateway, and a Streamlit dashboard.

PythonFastAPIvLLMNVIDIA NIMNVIDIA NeMoStreamlit

ShiftIQ

Open source
Reviewable, local-first code migration workflows
static analysis · dry runs · rollback checkpoints

Local-first code migration assistant for static analysis, dry-run migrations, MCP tools, rollback checkpoints, and PII/PHI/PCI-oriented pattern scanning.

PythonFastAPIASTMCPDry RunRollback

Annapurna-AI

Open source
Domain-specific agentic applications with model routing & cultural context
multimodal · model routing · cultural AI

Culture-aware, India-first AI meal planner and grocery assistant for South Indian vegetarian cooking with evidence-grounded recommendations.

Next.jsFastAPIGeminiLiteLLMSQLModel

Influence scales through people.

Mentoring, teaching, and making complex AI work easier to understand for practitioners across 20+ countries.

1,000+

Learners reached

20+

Countries represented

500+

Mentoring hours

70+

Topmate sessions

Anudeep is incredibly insightful, listening carefully and offering technical yet straightforward comments that are truly beneficial.

Michael

Freelance Developer

via ADPList
Extremely insightful and valuable discussion. Anudeep's depth of knowledge in Data and AI is evident, and his willingness to openly share his expertise is commendable.

Shashank H.V.

Student, UMass Dartmouth

via ADPList
His tailored advice on skills, job applications, and interviews was practical and insightful, leaving me confident and motivated.

Baran Khazaee

MSc CS, UC Davis

via ADPList
His strategic guidance and ability to simplify complex AI and career paths into clear, actionable steps were incredibly helpful.

Nelisa Sebastian

Data Analyst, Northeastern

via ADPList
An exceptional session, making complex Agentic AI concepts easy to understand. His motivating approach inspired me to take bold steps.

Mide Sowunmi

UX/UI Designer, Comcast

via ADPList
Contact / Plano, Texas

Bring me the problem that has to work in production.

Let's talk about production AI, platform architecture, mentoring, or research collaboration.