Location: Coimbatore or Chennai

Workplace Type: On-site

About Dataflo

Dataflo builds AI Products for US government agencies. Our products support public records management, legislative tracking, policy analysis, data intelligence, and AI-assisted workflows.

Public-sector software carries real consequences. The people using our products are accountable to citizens, auditors, courts, and elected officials. Outputs must be traceable, sensitive information must remain protected, and the product must work for staff who have little time for training.

“The feature mostly works” is not an acceptable standard.

The Role

As a Product Engineer, you will own a meaningful part of the product from beginning to end: understanding the problem, designing the experience, building the solution, deploying it, and verifying that it works in production.

You will work across product, design, engineering, data, and AI without treating any of them as someone else’s responsibility. The playbook for deploying AI products into government agencies is still being written, and you will help define it.

What You’ll Do

- Take customer problems from discovery through design, development, deployment, and post-launch verification.
- Build information-dense workflows, dashboards, and AI assistant interfaces that agency staff can understand quickly.
- Ship production code across React, TypeScript, backend services, REST APIs, authentication, SQL, integrations, and data pipelines.
- Build AI features using prompting, retrieval, structured outputs, evaluation frameworks, and human-review safeguards.
- Make trust visible by showing sources, confidence, decision history, and where human approval is required.
- Build accessibility into the product from the beginning, following WCAG and Section 508 requirements.
- Design secure workflows that protect personal information, enforce access controls, and create complete audit trails.
- Join customer discovery calls, review recordings, observe real usage,



and turn what you learn into shipped improvements.
- Decide what should remain specific to one agency and what should become a reusable product capability.
- Diagnose production issues, deploy fixes, verify them directly, and communicate clearly about what happened.
- Use AI development tools such as Claude Code, Codex, Cursor, or similar tools throughout planning, implementation, testing, and review.
- Build evaluation harnesses for AI-powered behavior so improvements and regressions are measured before release.

How We Work

- Own the whole loop. The person who finds a problem helps drive it through root cause, resolution, deployment, and verification.
- Move with urgency and judgment. We value fast execution, but speed does not replace security, testing, or careful review.
- Done means proven. A feature is not complete because the code was pushed. It must be deployed, tested in the real environment, and observed long enough to trust.
- Security and compliance are part of the product. The safe outcome should be the default. Sensitive actions require appropriate controls, uncertain outcomes stop for human review, and important activity must be auditable.
- AI-assisted, engineer-owned. AI tools can draft, explore, test, and accelerate the work. You remain responsible for understanding, reviewing, measuring, and owning everything that ships.
- Truth over ego. If something failed, say so. If an assumption was wrong, change direction when the evidence arrives.
- Meetings should create movement. Come prepared with evidence, options, and a recommendation. Leave with clear ownership and begin the work.





What We’re Looking For

- Experience shipping production software and taking real ownership of the outcome.
- Strong React and TypeScript skills, with attention to interaction details, loading behavior, errors, permissions, and edge cases.
- Comfort working with APIs, authentication, SQL, backend services, and messy real-world data.
- Practical experience building with LLMs, including prompting, retrieval, evaluation, and human-review workflows.
- Strong product and information-design instincts. You care whether the interface is understandable, not only whether it functions.
- The ability to turn an ambiguous problem into something useful that can ship quickly.
- A habit of reproducing issues, investigating evidence, and verifying fixes yourself.
- Clear written and verbal communication.
- Willingness to occasionally overlap with US working hours for customer discussions, launches, or urgent production issues.
- A portfolio or project history showing what you personally designed, built, and shipped.

Nice to Have

- Exposure to government technology, civic technology, healthcare, finance, or another regulated workplace.
- Experience with document-heavy AI products and retrieval over long, inconsistent source material.
- Experience designing data visualizations for non-technical users.
- Familiarity with audit logging, role-based access, data privacy, and compliance-oriented product development.
- Experience building accessible products to WCAG or Section 508 standards.

How to Apply

- Your latest resume and LinkedIn profile.
- Add this as a section in your resume: Submit your AI projects or prototypes you have personally built. It can be an AI product, prototype, internal tool, automation, workflow, or production feature that solved a real problem.

We care more about how clearly you understand the problem, the choices you made, and the outcome you achieved than about how polished the presentation is.

📌 Product Engineer, AI (Coimbatore)
🏢 Dataflo
📍 Coimbatore