Research Scientist, Applied AI in India at Jobgether

Own and advance the agent harness, including its core execution loop, tools, operating environment, guardrails, and model behavior.

Improve tool selection, context construction, coordination across sub-agents, and the handling of complex, long-running tasks through systematic experimentation.

Evaluate and select models for different tasks, balancing output quality, latency, and computational cost.

Design and implement evaluation frameworks for complex revenue-related tasks, including deal health, risk assessment, and pipeline analysis.

Develop evaluation datasets, benchmarks, grading methodologies, regression detection systems, and automated evaluation pipelines, particularly for tasks without clear ground truth.

Build predictive and scoring models using data from call transcripts, emails, and other business signals to identify cross-deal patterns and generate explainable risk indicators.

Lead model fine-tuning initiatives and advance the platform toward deeper model training and reinforcement learning.

Research and implement reinforcement learning approaches to improve agent performance and decision-making.

Strengthen context engineering and data quality through entity resolution, information extraction, and the development of knowledge and context graphs.

Take ownership of ambiguous research challenges from initial hypothesis and experimental design through training infrastructure, evaluation, and production deployment.

Collaborate closely with US-based AI leadership and the applied AI team in India to align research priorities with product goals.

At least 5 years of experience in applied AI research, applied science, or a closely related field, with the technical depth expected of a senior-level individual contributor.

A master’s degree or PhD in Computer Science, Mathematics, Statistics, or another quantitative discipline.

Hands-on experience building and improving agentic systems and agent harnesses, including tool use, multi-step reasoning, and context management.

Practical experience with large language models (LLMs) or large-scale deep learning systems, including model training, fine-tuning, and reinforcement learning from human feedback (RLHF).

A strong understanding of model behavior and the underlying principles that influence how models learn, reason, and respond.

Experience designing evaluation datasets and frameworks, particularly for complex tasks where reliable ground-truth labels are unavailable.

Demonstrated ability to independently define research directions, develop hypotheses, design rigorous experiments, and deliver results with minimal guidance.

Strong analytical, problem-solving, and experimental skills, with the ability to translate research findings into practical production improvements.

Excellent communication and collaboration skills, including the ability to work effectively with a US-based leadership team across time zones.

Publications in leading peer-reviewed conferences or journals are preferred.

Must-have skills: Agentic systems, agent harness design, and model fine-tuning.

Good-to-have skills: LLMs, applied AI, and AI research.

Competitive compensation: Annual salary range of INR 90–100 lakh (90,00,000–1,00,00,000).

Equity opportunity: Access to equity in an early-stage company ahead of its next funding round.

Remote flexibility: Work remotely from India, with opportunities for the team to meet in person in Bangalore.

Meaningful technical ownership: Lead the development of core agent capabilities from day one.

Direct leadership access: Work closely with the Head of AI and benefit from a close feedback loop with founders and product teams.

Collaborative environment: Join a focused, low-ego team that values speed, direct communication, and strong execution.

Research and innovation: Help shape an AI-first product as it advances from agent-harness development toward sophisticated fine-tuning and reinforcement learning.

Full-time opportunity: Take on a substantial research role with the potential to influence the product’s technical direction and long-term development.