Open to AI / LLM roles · Remote or relocation

Rahul Gond

>

AI Engineer with 3+ years shipping production LLM agents, RAG pipelines and Python backends. I build the AI for a trading-bot platform serving 20K+ users and 50K+ bots at Earnestroi, and I'm the founder of Dynavids, a live AI video-generation SaaS.

3+
Years in AI & backend
20K+
Users on platforms I build
50K+
Trading bots handled
10+
LLMs orchestrated in prod
// Building now

Scaling Dynavids: smarter scene planning, faster rendering and cheaper per-video model routing.

// At work

LLM agents and RAG for a trading-bot platform with 20K+ users and 50K+ bots.

// Looking for

AI / LLM / GenAI Engineer roles. Remote, hybrid or relocation, available for full-time.

// Featured · Founder

Building Dynavids from zero to live

An AI-native platform that turns a topic, PDF or URL into a narrated, motion-graphics explainer video, end to end.

LIVE · app.dynavids.com

AI Video Generation Platform

I designed and built the full product: the multi-model LLM pipeline, the async backend, credit billing, and the rendering engine. Videos up to 10 minutes, generated automatically or with human review at every stage.

Open App
Research
Script
Narration
Visuals
Scenes
Motion
Preview
Render

Multi-model routing

10+ LLMs via OpenRouter with per-stage model selection, quality profiles and automatic retries.

Cost-aware by design

Per-stage token and cost accounting feeding credit-based billing, tuned for a 60% gross-margin target.

Grounded & guarded

Ingests PDFs, images, websites and GitHub repos, with prompt-injection guardrails and schema-validated outputs.

Async job engine

FastAPI + Celery/Redis with persistent job state, queue isolation and stage-level regeneration.

Human-in-the-loop

Auto mode or review and approve each stage, so users control quality without starting over.

Programmatic motion

Reusable visual primitives, camera moves and layout validation, rendered with Playwright + FFmpeg.

FastAPICeleryRedisMongoDBOpenRouterReactCloudflare R2JWTPlaywrightFFmpeg
8pipeline stages
150+users
1,000+videos generated
≤10 minvideo length
// Experience

Where I've shipped

Founder & AI Engineer

Dynavids · Self-employed
Jun 2026 – Present
  • Built and launched an AI video-generation SaaS end to end: 8-stage LLM pipeline, async backend, billing and rendering.
  • Orchestrate 10+ LLMs with per-stage routing, cost tracking and guardrails; 150+ users, 1,000+ videos generated.
Generative AILLM OrchestrationFastAPICeleryReact

AI Engineer

Earnestroi Technology · Mumbai
Oct 2024 – Present
  • Build AI for a trading-bot platform (myaipro.ai) handling 20K+ users and 50K+ active bots.
  • Design LangGraph / LangChain agents for real-time market analysis with tool calling, structured outputs and stateful orchestration.
  • Implement RAG and vector-search pipelines that ground agent decisions in market and domain data.
  • Ship event-driven services on FastAPI, Redis, Celery and Kafka, plus risk automation (TP/SL, liquidation), monitored with Prometheus and Grafana.
LangGraphLangChainRAGVector SearchKafkaAWS

Python Engineer

Earnestroi Technology · Mumbai
Aug 2023 – Oct 2024
  • Built low-latency trading and backtesting engines in Python and FastAPI blockchain APIs.
  • Implemented Redis Pub/Sub real-time messaging and Dockerized deployments; prototyped early LLM agents and RAG.
PythonFastAPIMongoDBRedis Pub/SubDocker

Machine Learning Intern

Pianalytix · Remote
Jan 2023 – Jul 2023
  • Built classification and regression models with scikit-learn and XGBoost (EDA, feature engineering, tuning).
  • Developed a spaCy NER pipeline for support-ticket classification.
scikit-learnXGBoostspaCy
// Projects

Agents, RAG & real-time systems

Selected builds. Each one ships with a live demo or real users.

LLM Agent · Tool Use

AI DevOps Assistant

Turns plain-English DevOps requests into Bash commands and runs them on remote servers over SSH.

  • LangGraph agent with local models via Ollama
  • Command review and guardrails before execution
  • Microservice architecture with a Streamlit UI
LangGraphOllamaStreamlitSSH
Multi-Agent System

HireRoute: Multi-Agent Interview Prep

A technical-interview simulator where interviewer, evaluator and feedback agents run role-specific mock interviews.

  • Agents coordinate through shared LangGraph state
  • Structured scoring and actionable feedback
  • Redis sessions, Dockerized deployment
LangGraphLangChainRedisDocker
Production · 20K+ users

Smart AlgoTrading Platform

AI-augmented trading-bot platform with backtesting and execution engines and market-data ingestion, built at Earnestroi.

  • Handles 20K+ users and 50K+ active bots
  • Event-driven with Kafka and Redis; latency-optimized
  • LLM agents for market analysis, plus TP/SL risk automation
FastAPIKafkaRedisBacktesting
RAG · In progress

AI Document Q&A System

A document QA bot that answers with source citations and streams responses.

  • Semantic chunking with OpenAI embeddings
  • ChromaDB vector search and context assembly
  • Source-cited answers to reduce hallucinations
LangChainChromaDBOpenAIStreaming
LLM Agent · Tool Calling

Natural Language to Database Agent

Turns natural-language requests into validated CRUD operations through structured tool calls.

  • Separate parse, validate and execute nodes
  • Blocks unsafe or ambiguous writes
LangGraphFastAPIStreamlit
Founder · Live SaaS

Dynavids

AI video-generation platform with an 8-stage multi-model pipeline, credit billing and programmatic rendering.

  • 10+ LLMs with per-stage routing and cost tracking
  • Source ingestion from PDFs, web and GitHub
OpenRouterCeleryReactFFmpeg
// Open source

Latest on GitHub

All repos
// Stack

Tools I use in production

LLM & GenAI

LangGraphLangChainRAGAgentic AITool CallingStructured OutputsPrompt EngineeringOpenRouterLangSmithHugging FaceFine-tuning

Retrieval

EmbeddingsVector SearchChromaDBQdrantPineconeWeaviateSentence-Transformers

Backend

PythonFastAPIAsyncIOPydanticRESTWebSocketsJWT

Distributed & Data

CeleryKafkaRedis Pub/SubMongoDBSQLDaskRay

ML / DL

PyTorchTensorFlowKerasscikit-learnXGBoostspaCyOpenCV

Cloud & DevOps

DockerAWSCloudflare R2NginxGitHub ActionsLinuxPrometheusGrafana
// How I build

LLMs are easy to demo, hard to ship

These are the principles I use to turn model calls into software people can rely on.

01

Orchestrate, don't prompt

Stateful graphs with clear stages, typed state and structured outputs beat one giant prompt.

02

Ground everything

Retrieval, source citations and validation keep answers tied to real data.

03

Design for failure

Retries, persistent jobs, guardrails and human-in-the-loop checkpoints.

04

Watch the cost

Per-stage model routing and token accounting so every request is profitable.

// Education

Foundations

B.E. in Information Technology

Alamuri Ratnamala Institute of Engineering & Technology, Mumbai · 2023

Diploma in Information Technology

Thakur Polytechnic, Mumbai · 2020

// Contact

Let's build something intelligent

Hiring for an AI, LLM or Python backend role? I usually reply within a day.

Email me Resume
Email copied