Senior Applied Scientist

Senior Applied Scientist, Context Graph Team

MetAntz connects top 1% talent to global opportunities. Currently MetAntz is working with a Bay Area headquartered, niche IT service provider, working with several innovative companies for the Network Stack Developer role in Bangalore. The exciting work will be at our customer's customer site.

Senior Applied Scientist – Process Discovery & Agentic AI

About The Role

Before an AI agent can reliably operate software on a user's behalf, it needs to understand the underlying business process, including its steps, decision points, dependencies, variations, and edge cases. This knowledge is often scattered across event logs, unstructured documents, screen recordings, UI interaction traces, and domain expertise.

As a Senior Applied Scientist, you will provide technical leadership in developing scalable, process-agnostic learning systems that uncover the underlying structure, decision logic, and dynamics of complex workflows from sparse, noisy, and diverse observations.

You will own major research and technical areas, transforming heterogeneous data such as action logs, screen traces, semi-structured documents, and complex workflows into structured representations that support reliable AI agent execution. You will shape system architecture and define evaluation standards to ensure these representations are accurate, transparent, compact, and auditable.

You will also contribute to the development of computer-use models that enable AI agents to interact reliably with enterprise software.

This role involves solving open-ended research problems, developing new algorithms, designing rigorous experiments, making architecture decisions, and mentoring scientists and engineers.

Key Responsibilities

Process Discovery and Representation

• Lead the development of pipelines that discover and formalize end-to-end business processes from event logs, UI interaction traces, screen recordings, and unstructured documents.
• Develop algorithms for process variant detection, noise removal, and trace consolidation.
• Design core representations and architectures that preserve essential process information while eliminating irrelevant variations.
• Establish principles for balancing abstraction and fidelity in process representations.
• Build retrieval and grounding systems using standard operating procedures, documents, and domain knowledge.
• Define technical direction, review experimental results, and mentor scientists and engineers.

Evaluation and Reliability

• Develop rigorous evaluation frameworks for process discovery, including LLM-as-Judge pipelines, conformance checking, and regression testing.
• Create evaluation methods for processes without a single canonical ground truth.
• Test conditional branches, exception paths, loops, rare variants, and other complex scenarios.
• Make algorithmic and architectural trade-offs to support millions of execution traces and thousands of process variants.
• Diagnose production failures and translate recurring issues into research initiatives and system improvements.

Collaboration and Technical Leadership

• Collaborate with computer-use AI teams to ensure process representations support reliable autonomous execution.
• Work with research, engineering, and product teams to transition research prototypes into production systems.
• Mentor scientists and engineers in problem formulation, experimental design, technical reviews, and productionization.
• Contribute to technical documentation, research publications, and communication with technical and non-technical stakeholders.

Required Qualifications

• Strong Problem-Solving and Algorithmic Thinking: Ability to break down ambiguous problems, reason through edge cases, and design efficient algorithms using graph structures, data structures, approximation, indexing, and complexity analysis.
• Machine Learning Expertise: Deep knowledge of representation learning, sequence modeling, clustering, structured prediction, information retrieval, generative modeling, and foundation-model adaptation.
• Research and Technical Judgment: Ability to identify research opportunities, design meaningful experiments, evaluate competing approaches, and guide technical direction.
• Applied Foundation Model Experience: Hands-on experience building reliable production systems using modern language or multimodal models, including retrieval, structured generation, evaluation, grounding, and model adaptation.
• LLM Evaluation and Reliability: Experience with LLM-as-Judge systems, evaluation rubrics, synthetic data, fine-tuning, model post-training, confidence calibration, faithfulness verification, and production monitoring.
• Python Proficiency: Strong Python engineering skills, experience with ML frameworks such as PyTorch, and the ability to build scalable, tested, production-quality research systems.
• Cross-Domain Research: Ability to apply and adapt research from AI, computer vision, language modeling, process mining, program synthesis, graph learning, and knowledge representation.
• Communication and Mentorship: Ability to communicate complex technical concepts, document architectural decisions, and improve the work of other scientists and engineers.

Preferred Experience

• Strong research track record, including publications in leading ML, AI, NLP, computer vision, reinforcement learning, or data mining conferences.
• Experience taking research concepts from publication through production deployment.
• Experience defining technical strategy, mentoring researchers or engineers, and influencing architecture across teams.
• Knowledge of process mining, event log analysis, process discovery, conformance checking, or process variant analysis.
• Experience extracting structured information from documents, including named entity recognition, relation extraction, and table extraction.
• Experience processing multimodal data such as screen recordings, UI screenshots, interaction traces, and text.
• Experience with foundation-model adaptation techniques such as supervised fine-tuning, LoRA, synthetic data, distillation, and reinforcement-learning-based post-training.

Ideal Candidate Profile

You enjoy uncovering hidden structures in complex data and building systems that can infer how unfamiliar processes work without predefined rules.

You are comfortable moving between algorithms, machine learning, graph structures, retrieval systems, and foundation models to solve difficult research problems. You take ownership of technical decisions, value rigorous experimentation, and care about building reliable systems that work at scale.

You are motivated by working with an ambitious team where research, engineering, and practical execution come together to build next-generation AI systems.