Experience

My professional journey building production AI systems and data platforms.

Software Engineer

2 years
Infraspec
Bengaluru, Karnataka, India

February 2024 - Present

EdTech Client (US-based)

  • Designed and shipped production AI systems for student writing workflows, owning applied LLM architecture (prompt design, chaining, context management), evaluation, and full backend implementation.
  • Built an agentic essay-evaluation system generating rubric-aligned scores and guided feedback, automating large parts of teacher review while preserving assessment consistency.
  • Implemented reusable, multi-step LLM workflows to guide students through structured writing tasks, including parallel execution, input/output splitting, and aggregation.
  • Led model selection and migration by building a task-specific evaluation framework and cost-aware pipelines (batching, parallelism), moving production features from Anthropic Sonnet to LLaMA 4 Maverick.
  • Developed an internal dSAT question-generation system supporting configurable skills, difficulty levels, subjects, and subtopics, enabling scalable and consistent question creation.
  • Worked closely with domain experts, designers, and engineers to validate outputs and integrate AI features into the live platform.
LLM DesignPythonPrompt EngineeringEvaluationModel Migration

Fintech Client (Large Indian Brokerage)

  • Owned and optimized the Databricks data platform, enabling self-serve analytics workflows and reducing operational and ticket-driven support across teams.
  • Led a large-scale migration from Hive Metastore to Unity Catalog, migrating ~2,000 tables (~95 TB), improving governance, and cleaning up unused jobs and data to reduce costs.
  • Standardized data access and compute management processes, improving consistency and reducing onboarding friction for analytics and data science teams.
  • Built a reusable DS/ML pipeline template with CI/CD, observability, and code-quality best practices, reducing deployment time from multiple days to a few hours.
  • Developed backend systems to expose ML models via APIs and created a Databricks-based data tagging system for KYC image-validation model training under tight timelines.
DatabricksUnity CatalogData EngineeringML PipelinesPython

Intern

3 months
GRU
Bengaluru, Karnataka, India

October 2023 - December 2023

  • Applied Test-Driven Development (TDD) principles, implementing the Red-Green-Refactor cycle, Single Responsibility Principle and upholding the 3 Rules of Simple Design.
  • Gained expertise in testing methodologies, emphasizing the importance of testing, with hands-on experience in mocking and spying.
  • Led the development of an efficient report generation workflow, showcasing a practical application of acquired skills.
  • Used technologies like Django, Celery, PostgreSQL and SuperAGI.
  • Made a valuable contribution to SuperAGI by fixing a bug related to agent pausing during rate limit or try-again errors.
DjangoTDDCeleryPostgreSQLSuperAGI