Backend engineer building the systems behind production AI.

• Case study 01

  Care-home management platform

  An AI-assisted care-notes app and an email-driven referrals portal for UK child care homes, built and run end to end for a startup client.

  Pilot → production

  a one-month pilot, then about a year as the live product

  DecisionTook it from first requirements to production, owning the architecture, cloud and security throughout.

  FastAPIAuth0PostgreSQLRedis

  Read the case study →
• Case study 02

  SQL chat agent

  Streaming, tracing, load balancing, prompt management and an evaluation suite for an existing text-to-SQL chatbot.

  ~20k → ~7k

  base system-prompt tokens, tool results not counted

  DecisionMoved the schema out of the prompt and into a tool the agent calls, then proved it with evals.

  Azure AI Foundry agentsAzure OpenAIAzure API ManagementAzure DevOps

  Read the case study →
• Case study 03

  Citation-backed RAG chatbot

  A retrieval-augmented chatbot over SharePoint content with real-time text and speech, built for a US national website.

  Hybrid

  semantic, hybrid and keyword retrieval

  DecisionHybrid retrieval with citations, so every answer links back to the document it came from.

  Azure AI SearchAzure OpenAIWebSocketsVapi

  Read the case study →

1. Aug 2024 – now

   Backend Engineer · Newtuple Technologies

   Pune

2. Oct 2022 – Aug 2024

   Python Developer · Freelance

   Remote

3. 2019 – 2023

   B.Tech · Government College of Engineering

   Aurangabad

Strengths

What you get working with me.

01End-to-end delivery

From architecture to deployment: service design, authentication, CI/CD, networking and the WAF in front of it all. I take work from an idea to something running in production, not one slice of it.

02High ownership

I stay with what I ship. On the care-home platform I carried a product from pilot to production, and I still handle the deployments, maintain the evaluation suites and fix gaps I notice before anyone has to ask.

03Deep technical understanding

I learn the technology I work with well enough to explain its trade-offs: why a schema belongs in a tool and not in a prompt, how retrieval is scored, where a stream can break.

04Talking to clients

I take part directly in client conversations to understand what they need, then explain technical decisions and constraints in plain language.

05Evidence over opinion

Prompts and models change behaviour quietly. I build golden sets and judges so a change shows up as a number before it ships.