Mar 2026 – Present · 3 mos

• ● Lead the AI engineering organization building the company's production agentic AI platform for life-sciences commercial teams. The system autonomously plans and executes multi-step analytical workflows over customer data, delivering finished analytical artifacts (narrated insights, charts, and formatted reports) rather than raw answers, with multi-turn context continuity.
• ● Own the distributed-systems architecture underpinning the platform: a gateway–worker design with sticky-session routing over consistent hashing, autoscaling worker pools, service discovery, and full OpenTelemetry tracing. This enables predictable latency, fault isolation, and state continuity for long-running agentic conversations under production load.
• ● Architected a single-tenant external deployment model where each enterprise customer receives a dedicated, isolated instance exposed as an MCP server, with client identity fixed at boot and scoped skill and data allowlists. The same platform codebase serves multiple customers under strict data-residency and regulatory requirements.
• ● Designed a modular skills framework that lets the agent take bounded, auditable actions across retrieval, analysis, artifact generation, and external-system integration, with per-role and per-customer capability gating delivered via configuration rather than code releases.
• ● Established a production LLM evaluation system as the governance backbone of the platform: systematic response-quality measurement, automated validation pipelines, full traceability from answer to source data, prompt versioning, and per-query cost tracking. This is a prerequisite for trustworthy agentic AI in a regulated environment.
• ● Partner with product and executive leadership on commercial positioning and external-delivery readiness, aligning architecture, compliance posture, and team capacity with the sales pipeline.

Dec 2025 – Mar 2026 · 3 mos

• ● Own the next generation time series forecasting platform, leading design of hierarchical reconciliation pipelines that align forecasts across brands, markets, and channels while running reliably at large scale with clear SLAs, observability, and cost guardrails.
• ● Treated developer experience as a first-class responsibility, driving improvements in monorepo tooling, CI pipelines, internal CLIs, and documentation so that ML engineers, data scientists, and platform engineers could ship new models and LLM-powered features faster with fewer deployment issues and much smoother onboarding.

Jan 2024 – Dec 2025 · 1 yr 11 mos

• ● Managed two cross-functional teams, overseeing comprehensive project planning, execution, and delivery while actively fostering a culture of collaboration and continuous improvement to enhance team productivity and drive innovation across the organization.
• ● Spearheaded development and implementation of a centralized feature store as a feature generation source for ML models, ensuring complete feature traceability and significantly reducing costs by eliminating unnecessary data generation. The system enabled faster predictions/inference, facilitated rapid model iterations for the data science team.
• ● Led the team in mastering and implementing Dagster on Kubernetes for workflow orchestration, substantially enhancing feature store pipeline reliability and maintainability while ensuring cloud-agnostic capabilities for future scalability, which can be deployed in customers' cloud end-to-end.
• ● Directed the design and deployment of Omni-Channel Attribution and Attribution Sequencing models to enhance marketing attribution accuracy. Developed algorithms to determine optimal sequences of channels that maximize specific signals throughout the system over time, leading to substantial improvements in attribution analysis precision.
• ● Optimized complex data pipelines handling large volumes of data through strategically deploying tiered infrastructure and systematically resolving out-of-memory (OOM) and thresholding issues. Successfully increased resource utilization for a workload of approximately 5,000 container executions per hour (at peak), resulting in significant operational cost reductions.
• ● Designed and deployed an insight pipeline using Large Language Models (LLMs) for time-series prediction in the pharma sector, enhancing model explainability and decision-making processes.
• ● Led the strategic integration of Snowflake as a comprehensive data warehousing solution for analytics and platform teams, successfully transitioning from legacy systems.

Jul 2023 – Jan 2024 · 6 mos

• ● Cultivated a robust culture of collaboration and continuous improvement by initiating regular knowledge-sharing sessions and implementing streamlined communication processes, resulting in streamlined delivery cycles and accelerated innovation cycles.
• ● Spearheaded the implementation and supervision of comprehensive testing protocols, including expanded unit and integration testing frameworks. This initiative ensured the development of robust and reliable code, significantly elevating overall code quality standards.
• ● Led the development of a Pydantic-validated interface for unified configuration, dramatically improving consistency and effectively mitigating configuration issues from multiple sources, resulting in enhanced system reliability and reduced debugging time.
• ● Directed the strategic shift from AWS RDS dependencies to a more cost-effective data lake structure on S3, leveraging technologies like DuckDB and Polars. This move reduced operational costs and significantly improved the performance and time efficiency of critical project pipelines.
• ● Implemented a SOC2-compliant image strategy utilizing slim containers, successfully addressed cold-start issues, and optimized deployment efficiency across the platform. Introduced an advanced dependency management system with Poetry and Pyenv, ensuring consistently reproducible environments and streamlining the environment setup process.
• ● Integrated static code analysis tools and linters (Ruff, Yapf) into the CI/CD pipeline as pre-commit steps, enforcing stringent code quality standards and consistency. Additionally, optimized test infrastructure and environment setup processes, ensuring highly efficient and reliable testing integrated seamlessly into PR and pre-commit workflows.

Feb 2023 – Jul 2023 · 5 mos

• ● Successfully transitioned the organization to a monorepo structure, consolidating multiple repositories into one unified system to streamline machine learning code management and deployment processes, significantly improving development efficiency.
• ● Implemented comprehensive branch-specific tests and infrastructure deployment protocols for rigorous code validation, ensuring exceptional quality and stability of the machine learning codebase throughout the development lifecycle.
• ● Applied infrastructure as code principles across the entire pipeline, enabling version-controlled resource configuration and highly efficient management of machine learning infrastructure components.
• ● Designed and implemented an intuitive CLI (Command Line Interface) for the monorepo, effectively automating routine tasks, substantially reducing operational time, and minimizing potential errors in machine learning operations.
• ● Developed a general interface for unified configuration across all modules, ensuring consistent and thoroughly validated configuration settings throughout the machine learning pipeline, significantly reducing configuration-related issues and enhancing overall system reliability for different clients.