Headline

Give me enough tokens and the resources to use them, and I shall move the world. | Founding AI/ML Engineer | Multimodal Agents · LLM Evals · Production AI

About

I'm the founding engineer at GoReach Finance, where I built the AI and data backbone of a SEBI-regulated mutual fund advisory platform from zero: the conversational AI service, the realtime spine that connects it to the backend and the customer app, the fund data layer, portfolio analytics, and notifications. The AI layer routes across Claude, Gemini, and OpenAI, with cost controls and evals that gate every output before it reaches a customer. This is regulated advice, so "the model said so" doesn't cut it. The compliance gates are code, they have tests, and you can watch them fire. Around that: fund search and portfolio analytics, market data, exchange integration for order placement, Aadhaar-based KYC, the React Native customer app, and a Hinglish speech pipeline that was harder than the rest of it put together. Java and Spring Boot on PostgreSQL, Python for the AI service, AWS underneath. Before Reach I was the founding AI/ML engineer at JuiceLabs, where I built the idea-to-video generation pipeline: a multi-phase agentic system with LLM-as-judge evaluation from a different model family and user approval gates in front of every expensive step. It shipped ads at millions of impressions a month for enterprise clients. Before that I was the founding engineer at Sytolab, building the human-machine interface and the OpenCV quality control layer for a biomedical 3D printer used in organ regeneration research. I also ran weekly Claude Code sessions for a 20-person engineering team for four months, and I consult with enterprise teams on getting real work out of coding agents. In my experience the bottleneck is rarely model access. It's what happens to the output after the agent hands it over.

Location

Bengaluru

AI stack

anthropic apis

claude code

codex

hermes agent

ChatGPT

Veo

Work experience

Founding AI/ML Engineer at GoReach, Jul 2026 - Present · 3 mo, Bengaluru, Karnataka, India, - Built the AI and data backbone of a SEBI-regulated mutual fund advisory platform from zero: the AI service, the backend, the customer app, and the infrastructure under them. - The AI layer routes across Claude, Gemini, and OpenAI, with cost controls and an eval suite that gates every output for compliance, PII, and scope before it reaches a customer. - Regulated advice can't rest on "the model said so," so those gates are code and they have tests. - The backend covers fund search and portfolio analytics, market data, exchange integration for order placement, and Aadhaar-based KYC. The customer app is React Native, built end to end. - Speech was the messy one: live Hinglish transcription, which is harder than either language on its own. Java and Spring Boot on PostgreSQL, Python for the AI service, AWS underneath.

Founding AI/ML Engineer. at JuiceLabs AI, Oct 2025 - Jul 2026 · 10 mos, Bengaluru, Karnataka, India, - Built JuiceLabs' video generation pipeline — idea to final cinematic video, length-agnostic, shipping ads at millions of impressions a month for enterprise clients. Multi-phase agentic architecture, parallel multi-model orchestration across Opus 4.6, gpt-image-2, Seedream 4.5, and Seedance 2.0. Cross-family LLM-as-judge evaluation so the generator and grader don't collude. User-approval gates in front of every expensive step. - As the founding AI/ML engineer: data ingestion, prompt steering across multimodal models, agent workflows, automated evaluation, and production inference tooling. FastAPI + Firestore + GCP backbone, LangFuse for observability, Qdrant for hybrid dense + sparse retrieval (Vertex text-embedding-005 + SPLADE), direct SDK integrations with OpenAI, Anthropic, and Google. - Wrote a unified LLM router across 6 providers (OpenAI, Anthropic, Google Vertex AI, OpenRouter, xAI, Perplexity) — per-model backup chains, Anthropic ephemeral prompt caching, extended thinking budgets, provider-specific parameter cleanup on fallback. - Ran weekly Claude Code sessions for our 20-person engineering team. Four months total. Covered prompt strategies, agentic workflow design, and how to integrate Claude into actual day-to-day engineering work.

Founding Software Engineer at Sytolab, Jan 2024 - Oct 2025 · 1 yr 10 mos, Bengaluru, Karnataka, India, - Founding software engineer. Built Syto Studio, the human-machine interface for biomedical 3D printers. React frontend, Flask backend. Integrated Klipper API for STL uploads, slicing, and temperature controls; Moonraker for real-time printer monitoring. - Researchers used the system to print tissue constructs placed into living animals for organ regeneration studies. - Wrote the OpenCV-based computer vision layer for real-time print quality control. Reduced user setup time by 30% and improved print accuracy by 50%.

Software Development Engineer in Test at Paytm, May 2023 - Jan 2024 · 9 mos, Remote, - Automated device testing using Python, reducing testing cycles by 50% and eliminating 25 hours/week of manual effort. - Authored 100+ manual test cases for Soundboxes, covering functional, regression, and integration scenarios. - Created comprehensive test plans and maintained documentation for end-to-end testing cycles. - Integrated AWS services (S3, DynamoDB) with Boto3 SDK to streamline test data management. - Generated Matplotlib-powered test reports, identifying 50+ critical bugs pre-release and cutting post-release issues by 20%. - Validated Kafka-based real-time data pipelines and REST APIs (Postman) for 10M+ transactions. - Leveraged JIRA to track and resolve 50+ critical bugs, collaborating with cross-functional teams to prioritize fixes. - Conducted SQL queries to validate transactional data stored in DynamoDB, ensuring consistency across 10M+ transactions. - Integrated automated tests into Jenkins CI/CD pipelines, reducing deployment delays by 15%.

QA Automation Intern at Paytm, Jan 2023 - May 2023 · 5 mos, - Developed and implemented an automated testing framework that streamlined the testing process, saving the team over 50 hours per week. - Collaborated with business stakeholders to understand requirements, leading to the development over 100 automated test cases. - Performed API testing using Postman

Product Analyst at Terribly Tiny Tales, Oct 2021 - Jan 2023 · 1 yr 4 mos, - Revamped frontend UI/UX using React and Cypress, boosting website traffic by 20% and reducing production bugs by 40%. - Collaborated with UX designers and product managers to align features with customer requirements. - Collaborated with DevOps to streamline microservices testing in Docker containers, improving deployment reliability. - Performed cross-browser testing (Chrome, Firefox, Safari) for React-based web applications, resolving 30+ UI compatibility issues. - Automated regression testing with Cypress, increasing test coverage by 35%.

Education

Bachelor of Technology - BTech hons, Electrical, Electronics and Communications Engineering, Dr. A.P.J. Abdul Kalam Technical University

St Dominic Savio College

Public links

github: https://github.com/dhruvpuri

dhruvpuri.com: https://dhruvpuri.com/