Open to AI Engineering roles

Hi, I’m KuldeepAI Engineer

I build AI systems that run in production, not notebooks that stop at the demo — retrieval-backed agents, Text-to-SQL over live databases, and models served behind real APIs.

Kuldeep Patra
See my work → Get in touch
4platforms shipped
2live in production
48hagent, end to end
MCAAI & ML, KIIT
PythonRAGHybrid RetrievalLlama-3 / GroqText-to-SQLXGBoostscikit-learnFlaskPostgreSQLReactTypeScriptPocketBasePrompt EngineeringAI SafetyPower BIPythonRAGHybrid RetrievalLlama-3 / GroqText-to-SQLXGBoostscikit-learnFlaskPostgreSQLReactTypeScriptPocketBasePrompt EngineeringAI SafetyPower BI
How I work

Understand, build, then verify

Most of what I have learned came from the parts that went wrong, so the last step is the one I refuse to skip.

01

Understand

Start with the constraint that actually bites — unreliable internet at a restaurant till, a knowledge base that must not invent prices. The constraint decides the architecture.

02

Build

Enforce the rules where they cannot be bypassed. Business invariants in the database, not the UI. Retrieved context treated as untrusted input, not instructions.

03

Verify

Check retrieval before judging generation. Cross-validate before quoting a metric. That habit is how I found target leakage that had inflated a model’s score.

Selected works

Things I have built and shipped

Four projects, each with a full case study — the problem, the decisions I would defend in an interview, and the limitations I would not hide.

Live demoAI Patient Intake Agent

AI Patient Intake Agent

Diagnosed why a naive crawl returned marketing pages instead of clinical content — and fixed retrieval at the source.

RAGHybrid retrievalSalesforcePrompt engineering
View case study →
In productionMulti-Restaurant POS

Multi-Restaurant POS

Database-per-tenant, so a restaurant keeps trading when its internet drops.

React 19TypeScriptTanStackPocketBase
View case study →
In productionHotel Property Management System

Hotel Property Management System

Replaced a clipboard-and-paper-slips process with live state every role reads from the same place.

ReactTypeScriptPocketBaseRBAC
View case study →
ML + LLMHR Attrition AI Platform

HR Attrition AI Platform

0.805 ROC-AUC on 5-fold CV — and a target-leakage bug found and documented rather than quietly benefited from.

PythonXGBoostscikit-learnFlask
View case study →
Kuldeep Patra
Who I am

Building at the application layer of AI

An engineer who enjoys the intersection where a model stops being a notebook and starts being something a business relies on daily. A naive crawl that indexed marketing pages. A prompt-injection vector sitting in source data. Target leakage that made a model look far better than it was. Those are the sections I write up most carefully, because an engineer who can only describe the happy path has not finished the work.

  • Based inBengaluru, India
  • Independent workSince 2025
  • DegreeMCA, AI & ML · KIIT
Journey

Education and independent work

A business degree first, then a master’s in AI and machine learning, with freelance production work running alongside it.

  1. 2025 → Now

    Freelance & independent projects

    Solo, end to end

    A multi-outlet restaurant POS and a hotel property management system, both live in production, plus an e-commerce analytics platform and a website rebuild for an international client.

    RequirementsArchitectureDeploymentProduction support
  2. 2024 → 2026

    Master of Computer Applications

    KIIT University, Bhubaneswar

    Artificial IntelligenceMachine Learning
  3. 2021 → 2024

    Bachelor of Business Administration

    Samanta Chandra Sekhar College, Puri

Certifications

Credentials and coursework

Structured coursework that runs alongside the build work.

Verified

Excel Basics for Data Analysis

IBM · Coursera

Jul 2026Verify

Google Data Analytics

Google · Coursera

2026

Artificial Intelligence & Machine Learning

CTTC Bhubaneswar

2025
In progress

Machine Learning Specialization

Stanford & DeepLearning.AI

Ongoing
Toolkit

What I work with

Grouped by what it is for. Everything here appears in the work above.

AI & LLM

RAG knowledge basesHybrid keyword + vector retrieval LLM API integrationLlama-3 via GroqText-to-SQL agents Prompt engineeringPrompt-injection mitigationAI safety guardrails

Machine learning

Pythonscikit-learnXGBoostpandas NumPySHAPFeature engineeringModel evaluation

Backend & data

FlaskREST API designPostgreSQLSQL psycopg2 / SQLAlchemyPocketBaseRole-based auth

Frontend & reporting

ReactTypeScriptJavaScript TanStack Router / QueryTailwind CSSPower BIGit
Let’s work together

Let’s make it happen

Open to AI Engineering roles — LLM applications, retrieval systems and the engineering around them.

kuldeeppatra8@gmail.com LinkedIn GitHub