Suresh.saragadam
LinkedInResume

AI Engineer · Visakhapatnam

Suresh Saragadam

Senior engineer with ~10 years across React and Node.js. For the last 1.5+ years I've been shipping a production GenAI assistant at NielsenIQ that understands user intent and answers from real product data.

Intent-aware LLM assistants · grounded responses · eval-safe prompt shipping.

Let's talk
LinkedInGitHubResume
Intent-aware LLM assistantsAzure OpenAI · Python · PydanticPrompt evals · token optimizationReact + Node.js
Suresh Saragadam, AI Engineer

React · Node.js · Azure OpenAI · Python

Selected work

From full-stack product work to production GenAI assistants.

Years building React experiences and Node/API services — now applying that craft to intent-aware bots, grounded answers, and eval-backed prompt changes.

NielsenIQ · in production

1.5+ years of GenAI work

Intent-aware analytics assistant

A production conversational assistant for retail analytics. It interprets what the user is asking, maps that intent to data and responses in our product, and returns answers teams can trust. Behind the calm UX: Azure OpenAI orchestration, Pydantic-structured pipelines, token-conscious prompting, evaluator suites, and Grafana-backed monitoring so prompt changes do not silently break existing flows.

Azure OpenAIPythonPydanticPrompt evaluationToken optimizationGrafanaReactNode.js

Schemax / Brandix

Major project

Brandix apparel manufacturing platform (v2)

Mono-repo (Nx/NRWL) platform spanning decentralized codebases across 36 plants for shop-floor visibility and productivity tracking.

Nx / NRWLREST APIsMulti-plant architecture

Oct 2022 — Present · 4 yrs 1 mo

Senior Software Development Engineer

NielsenIQ (NIQ)

Active now
  • Building a production GenAI assistant for retail analytics (small and medium CPG manufacturers) that interprets user intent and answers from data in our systems.
  • Integrated Azure OpenAI with agent-style tool calling and multi-step response flows; used Python and Pydantic for structured outputs and reliable pipelines.
  • Kept prompt iteration safe with automated evaluators; optimized token usage for cost and latency; monitored LLM and service health with Grafana.
  • Worked across the stack — Python for LLM services, React for product UX, Node.js where the platform needs it — and shipped reusable UI components adopted across teams.

Jan 2021 — Sep 2022 · 1 yr 8 mos

Senior Software Development Engineer

Tata Consultancy Services (TCS)

  • Delivered React-based retail analytics experiences for CPG manufacturers.
  • Built reusable frontend components that accelerated feature delivery across modules.
  • Recognized with Innovista Award and client Duct Tape Award for module delivery impact.

Feb 2017 — Dec 2020 · 3 yrs 10 mos

Software Development Engineer (joined as intern)

Schemax Expert Techno Crafts

  • Built Shop Floor Control APIs for Brandix on an Nx mono-repo platform spanning 36 plants, so plants could track garment manufacturing productivity.
  • Supported BIEAP public examination paper valuation systems for Andhra Pradesh — triage and bug fixes under live constraints.
  • Integrated Keycloak for centralized user management across modules.
  • Delivered vehicle service lifecycle tracking with IoT collaboration and payment gateway integration.

Open-source project

AI Explorer

One product, many parts. Chat, prompts, streaming, history, tools, SQL, embeddings, RAG, agents, LangGraph, MCP (real server + client), guardrails, Langfuse hooks, Prometheus, Grafana, Redis, Docker, CI/CD, and evals — without restarting a new demo each time.

FastAPIPydanticOpenAINext.jsFAISSLangGraphDockerGrafana
View on GitHubAll projects

Capabilities

React up front, Node underneath, LLMs in the loop.

Full-stack product engineering plus production GenAI — intent understanding, Azure OpenAI, structured outputs, token-aware prompting, and evaluators that keep prompt changes safe.

AI / LLM

  • Intent understanding & grounded responses
  • Azure OpenAI (GPT-4o / mini / o-series)
  • Agent-style tool calling & multi-step flows
  • Prompt evaluation & regression tests
  • Token optimization (cost & latency)
  • Python · Pydantic structured outputs

Frontend

  • React
  • TypeScript
  • JavaScript
  • Analytics UI
  • Reusable component libraries

Backend

  • Node.js
  • Python
  • NestJS
  • REST APIs

Data

  • PostgreSQL
  • MySQL
  • MongoDB

Platform

  • Grafana monitoring
  • Docker
  • Nginx
  • Keycloak
  • Linux
  • GitHub

About

Frontend, backend, and production LLM systems.

The assistant looks simple on the surface and is hard in production. The real work is reliability: making sure a prompt change doesn't silently break existing flows, keeping token cost and latency in check, and knowing when quality drifts. I've tackled that with Azure OpenAI, Python, Pydantic structured outputs, automated evaluators, and Grafana monitoring.

Earlier chapters spanned Brandix manufacturing systems at Schemax and retail analytics at TCS and NielsenIQ. That mix of React UIs and Node/API work is what I now lean on when shipping assistants that have to understand intent and answer from real product data.

Education

B.Sc. in Computer Science

Dr. Hima Sekhar Degree College · Visakhapatnam, India

2012 — 2015

Quick facts

  • Based in Visakhapatnam, India
  • ~10 years shipping software, React and Node.js throughout
  • 1.5+ years on production GenAI at NielsenIQ

Recognition

  • Innovista Award (TCS) for team contribution in module development.
  • Client Duct Tape Award (TCS) for pragmatic delivery under pressure.
  • Excellence Award at Schemax for extraordinary quarterly performance.
  • USA client appreciation for delivering a project in one week on a new technology stack.
  • CTO appreciation (Schemax) for ramping on new technologies quickly.

Contact

Open to AI Engineer conversations.

Happy to talk about intent-aware assistants, evals, or full-stack GenAI product work.

saragadamsuresh174@gmail.comLinkedInGitHub