Senior Data Scientist

Voice AI for enterprise BFSI sits at two extremes: generic LLM wrappers that sound robotic, and legacy IVR systems that frustrate callers. Arrowhead exists because the most important conversations in financial services, loan sales, insurance, collections, deserve a voice that sounds and responds like a person.

That belief produced a fully in-house voice AI platform: small language models, fine-tuned ASR, and optimized TTS, all built to handle the complexity of real BFSI conversations at scale.

Arrowhead is Seed-funded and counts Aditya Birla, Axis Bank, Paytm, CRED, and Shriram Finance among its clients, led by a founding team with engineering backgrounds from Amazon, Airbnb, Uber, and Rippling.

As Senior Data Scientist at Arrowhead, you will own the technical direction of the ML models powering hundreds of thousands of BFSI voice calls daily.

You will come in deep on either LLM post-training and fine-tuning or ASR and TTS optimization, and grow into the other side over time. The role asks for the ability to identify the right bets, make sharp trade-offs, and unblock the people around you.

You will work directly with the CTO and operate as the senior technical voice on a three-person ML team.

Technical direction of the ML stack. Identify the highest-leverage optimizations across the voice AI pipeline and set the path the team runs on.

LLM post-training and fine-tuning. Own the small language models that drive conversational intelligence, from architecture decisions to deployment.

ASR and TTS optimization. Fine-tune and improve speech recognition and synthesis models for accuracy and naturalness at production scale.

End-to-end model development. Shepherd ML features from requirement and design through implementation, evaluation, and deployment.

ML infrastructure and pipelines. Build the training, evaluation, and experimentation systems the team runs experiments on.

Technical mentorship on the team. Guide the ML engineer and data scientist on trade-offs, priorities, and the right bets to take.

Fine-tuned or post-trained a language model. You have gone past inference and API calls; you have touched weights, run training jobs, and shipped the result.

Optimized ASR or TTS models in production. You have improved speech recognition accuracy or synthesis quality on real, noisy, high-volume data.

Set technical direction without a manager above you. You have identified what to build next and made the case for it, in an environment where the answer was not handed to you.

Unblocked teammates on hard problems. You have guided others through trade-offs and design decisions, as an IC, without a formal management title.

Operated in a fast-moving, ambiguous environment. Requirements shifted, priorities changed, and you moved with them without losing output quality.

Read research and shipped from it. You have taken a paper, evaluated its applicability, and implemented something from it that made it into a production system.

Own the problem end to end. Take a question from first principles to a shipped model; handoffs are the exception here.

Make the call on thin information. Decisions move fast; bring your reasoning, state your assumptions, and commit.

Raise the bar by doing, not directing. The senior technical voice here earns influence through the quality of the work, not the title.

Read the research, then build from it. Stay current with what is happening in NLP and speech; bring the relevant findings to the team before the team asks.

Work in the room. The team is on-site in Indiranagar, Bengaluru, five days a week; the best thinking happens in person.

Say what you see. Ideas are heard and acted on here; if you spot a better path, name it.

Cash compensation of ₹40L to ₹80L a year, scaled to experience. Equity is part of the package and structured flexibly based on the profile. You report directly to the CTO and operate as the senior technical voice on the ML team. The role gives you full ownership of the models behind one of the fastest-growing voice AI platforms in BFSI.