Applied AI / Research Engineer

Location: India, Remote

Type: Full-time

Team: AI / Engineering

About hpy

hpy is building AI that understands people, not just language.

Most of today’s AI systems are remarkably good at understanding what someone said. They are considerably less capable of understanding how someone is feeling, why their emotional state is changing, and what kind of interaction might actually improve their wellbeing.

We think that distinction matters.

Human wellbeing is influenced by patterns that unfold over time: emotion, behavior, relationships, environment, physiology, meaning and context. Today’s models see fragments of this picture. We want to help them understand more of it.

We started in behavioral health, building technology that understands conversations, identifies emotional signals, synthesizes longitudinal context and helps professionals better understand the people they serve.

But behavioral health is our starting point, not the boundary of what we’re building.

We are exploring the intersection of AI, emotion and human wellbeing and how hpy’s underlying intelligence can extend into entirely new products and verticals.

The question we’re interested in is

What becomes possible when AI can understand not only what a person is doing, but how they are doing?

We’re looking for an exceptional India-based engineer to help us answer that question.

The role

This is a hybrid engineering and applied research role.

You will work closely with hpy’s founder and our partners to extend hpy’s core intelligence into new products and verticals.

You should be comfortable moving across the entire stack: building an interface, designing the backend, constructing the AI pipeline behind it, evaluating a new model and putting the resulting product in front of a partner.

Some projects will begin with a customer problem. Others may begin with a new research paper, model or capability that makes something possible that wasn’t possible six months ago.

We expect you to be paying attention to both.

What you’ll work on

You will

• Extend hpy’s emotion and context intelligence into new wellbeing verticals.
• Build experiments around how models understand emotion, behavior and longitudinal human context.
• Prototype and ship full-stack AI products from idea through production.
• Evaluate frontier foundation models, multimodal models and emerging AI infrastructure.
• Build agentic systems that reason across conversations, structured information and external tools.
• Develop evaluation frameworks for measuring emotional and contextual understanding.
• Explore fine-tuning, representation learning, embeddings and domain-specific model adaptation.
• Work with large conversational and emotional datasets.
• Work directly with partners to deeply understand workflows and translate ambiguous problems into products.
• Move rapidly from research → prototype → user feedback → production.

What we’re looking for

We’re looking for an unusually strong generalist engineer with deep curiosity about AI.

You should be able to move comfortably between research and engineering. You don’t need someone else to translate a paper into an experiment or an experiment into a product.

We value

Exceptional technical foundations. Strong computer science fundamentals and the ability to work across frontend, backend, APIs, databases, infrastructure and ML systems.

Deep AI fluency. You actively follow frontier models, papers and tooling. When something important ships, you want to understand how it works, test it and determine whether it changes what we can build.

Strong product instincts. You care about the human problem rather than simply completing a specification.

Excellent communication. You can speak directly with partners, ask good questions, understand complex workflows and turn those conversations into technical decisions.

High agency. You are comfortable being given an outcome rather than a ticket.

Intellectual curiosity. You are interested not only in making models work, but understanding why they work and where they fail.

We have a strong preference for candidates from top-tier engineering and research universities in India or globally, or candidates who can demonstrate an equivalent level of technical ability through exceptional research, open-source work or products they have built.

Major plus

Experience developing or training models from scratch.

We are particularly interested in candidates who have gone deeper than simply calling foundation-model APIs.

Experience with any of the following is valuable

• PyTorch / JAX
• LLM training, fine-tuning and post-training
• Multimodal models
• Speech and conversational intelligence
• Emotion / affective computing
• Representation learning
• Reinforcement learning
• Model evaluation and interpretability
• Large-scale data pipelines
• Agentic systems and tool use
• Human-computer interaction
• Computational psychology or behavioral science

You don’t need all of these.

Depth, curiosity and the ability to learn quickly matter more than matching a checklist.

How we think about wellbeing

We don’t think wellbeing is an application category.

We think it is a fundamental human state.

People continuously move through emotional states shaped by relationships, work, health, sleep, environment, identity, purpose and thousands of small interactions.

Historically, software had almost no ability to understand this state.

AI changes that.

Models can increasingly listen, observe patterns, reason across long periods of time and integrate multiple forms of human context.

That creates an important technical and scientific question

Can machines develop a sufficiently rich understanding of human emotional context to help people live better lives?

We don’t know the complete answer.

We want to build the systems, experiments and products that help us find out.

Why hpy

This is not a role where you’ll spend six months implementing a predetermined roadmap.

You’ll work on problems that don’t yet have established solutions.

You’ll have access to real-world problems, real users and a growing body of emotional and conversational intelligence developed through our work in behavioral health.

You’ll work directly with the founder and have significant ownership over what gets researched, prototyped and ultimately shipped.

You’ll be expected to challenge assumptions, experiment aggressively and occasionally tell us that the thing we thought we should build is the wrong thing.

We’re looking for someone who wants the unusual combination of research, engineering, product creation and real-world deployment normally found at very early-stage companies and frontier AI labs.

If you’re obsessed with what models can understand about people, and you want to turn that curiosity into systems used in the real world, we’d like to talk.