Keval Padsala

Python developer · LLM systems and backend engineering

I don't just write APIs. I build the product around them.

I'm a Python engineer working on LLM systems. Give me a rough idea and I'll take it to production: the architecture, the integrations, the debugging. I think about the client's goal and the person using it before I think about endpoints.

Surat, India · GenAI Automation Engineer at Ameridenti LLC

Selected workResume (PDF)Get in touch

What I’ve built

  1. OwentraAI + human customer chat, sold as SaaSFounder, built end to endOwn product
  2. ZettalgorConversational AI platform and IDE assistantAPI layer, VS Code extensionIraGlobaltech
  3. IndicatorleadsTrading platform: manual, algorithmic, AIExecution and portfolio modulesIraGlobaltech
  4. callingagents.inRAG voice agent for business callsRAG pipeline, voice integrationIraGlobaltech
  5. MedendxClinical conversations to SOAP notesEnd-to-end LLM pipelineIraGlobaltech
  6. Crawling pipelinesScrapy crawlers and ETL, millions of recordsSpiders, ETL, deploymentBoth roles
  • LLM systems
  • Real-time
  • Data pipelines
  • Web platforms

Experience, live

Building production software since Jan 2024

Career

Work experience

Where I've worked and what I did there, most recent first.

  1. Sep 2026 – Present

    Current role

    GenAI Automation Engineer

    Ameridenti LLC

    • GenAI automation for a medical billing software company that automates US healthcare billing processes to US standards.
    • Automated claim scrubbing, alongside other tools that take manual steps out of the billing workflow.

    GenAI automation · Claim scrubbing · US medical billing

  2. Jan 2025 – Jul 2026

    Python Developer

    IraGlobaltech · Surat

    • Built LLM products and real-time systems for client platforms, usually starting from an open-ended request and owning architecture, integration and production debugging.
    • Designed and built the trade execution and portfolio modules of an AI-based trading platform, with live Binance data over WebSockets and Celery for background processing.
    • Designed an OpenAI-compatible API layer and built a VS Code extension for a conversational AI platform running on open-source LLMs.
    • Built a RAG voice agent with Twilio that books appointments on its own, and an LLM pipeline that turns clinical conversations into SOAP notes.

    Python · Django · Django REST Framework · Celery · WebSockets · LangChain · RAG · Ollama · Twilio · Scrapy

    Projects: Indicatorleads, Zettalgor, callingagents.in, Medendx, Crawling pipelines

  3. Jan 2024 – Jan 2025

    Python Developer

    Akruz Creative Pvt. Ltd. · Surat

    • Designed REST APIs and PostgreSQL data models for real-estate and e-commerce platforms: property listings, advanced search, user interactions and orders.
    • Implemented secure transaction workflows on a modular backend shared by both platforms, built for concurrent user load.
    • Engineered reusable Scrapy spiders that extracted millions of structured records, deployed on Zyte Cloud with scheduled runs and monitoring.

    Django · Django REST Framework · PostgreSQL · Scrapy · Zyte · Neon

    Projects: Akruz platforms, Crawling pipelines

Projects

Selected work

What each product does, my part in it, and the decision or problem that mattered most.

Owentra

LLM systemsOwn product · Founder

Live, profitable, with paying customersowentra.com

A live SaaS platform where a business trains an AI assistant on its own website content. Visitors get answers at any hour in the brand's voice, and staff can step into any conversation.

My roleFounder. I built the whole product, from concept to paying customers.

What I built

  • An embeddable chat widget, deployed with one line of code, combining appointment booking, scripted event-based flows and free-form AI chat.
  • A live agent console over WebSockets: staff see every conversation and take over with one click.
  • Tiered, usage-based pricing from Free to Business plans, integrated into the product.

Key decision

Hand off to a person instead of letting the AI guess

When the assistant can't resolve a query, the conversation passes to a human mid-chat, so the business keeps control of every customer interaction.

ResultLive and profitable, with paying customers.

Python · Django · Django templates · WebSockets · LLMs · PostgreSQL

What I bring

Capabilities

Grouped by area, with links to the work where each was used.

Product and user experience

I don't only code APIs. I work from the whole product: what the client is trying to achieve, how the frontend will use the backend, and what the person using it experiences.

Turning open-ended client requests into shipped products, Backend designed around the frontend and its users

See: Owentra, Zettalgor

LLM systems

RAG, LangChain, Prompt engineering, Ollama, Self-hosted open-source LLMs, OpenAI-compatible APIs

Lighter use: LangGraph, Hugging Face

See: Zettalgor, callingagents.in, Medendx

Backend and APIs

Python, Django, Django REST Framework, REST API design, PostgreSQL, Neon

Lighter use: Flask

See: Owentra, Akruz platforms

Real-time and async

WebSockets, Celery, Binance streaming API, Twilio speech-to-text and text-to-speech

See: Indicatorleads, Owentra

Day-to-day tools: Git, Postman, Claude Code, Antigravity

Credentials

Education and certifications

  • 2020 – 2024

    B.Tech, Computer Engineering

    Bhagwan Mahavir University, Surat

  • 2023

    Data Science and ML Intern

    UniConverge Technologies

    Deep-learning model for real-time crop and weed detection.

  • 2022

    SAP Certification, Industry 4.0 Technology

    Edunet Foundation

    Violence-detection project using machine learning and computer vision.