My two failed startups taught me what no MBA ever could

System Architecture & Computer-Vision Pipeline Leadership: Leading end-to-end technical architecture for an offline, on-premise computer-vision assistance application, defining module interfaces, data formats, and APIs across an engineering team; coordinating model selection for object, scene, pose, hand-position, and action recognition, and owning camera-to-application workflow design so vision outputs integrate cleanly with downstream routine and AI-agent modules.

Agentic AI & On-Premise LLM Integration Architecture: Architecting the integration layer between the computer-vision pipeline and the application's offline agentic AI and local LLM components, applying natural language processing for on-device prompt understanding, and defining event schemas and APIs so vision outputs feed directly into on-device retrieval-augmented reasoning and step-by-step application guidance, keeping the iOS/Android application a single coherent, fully on-premise AI system with no third-party cloud dependency.

Cross-Functional Leadership, Code Review & Technical Risk Management: Leading and mentoring a 3-person development team across full-stack integration, image-processing, and offline agentic-AI engineering; directing pull-request approval and integration review across all workstreams, resolving cross-module conflicts, preventing feature duplication, and reporting technical blockers and safety-relevant risks to keep the team aligned on shared architecture and standards.

Data Collection & Validation: Collected, validated, and cleaned institutional library datasets from multiple source systems to support operational reporting and usage analysis.

Exploratory Data Analysis: Conducted exploratory data analysis (EDA) to identify resource demand trends, usage patterns, and data quality gaps across services.

Data Visualization & Reporting: Developed visualizations and analytical summaries to communicate insights and support data-driven planning decisions.

Stakeholder Collaboration: Collaborated with stakeholders to translate reporting requirements into structured, analysis-ready datasets and actionable insights.

LLM Integration & RAG: Developed GPT-3.5–powered enterprise applications using Python (FastAPI), Azure OpenAI, and LangChain; designed modular prompt templates, system instructions, and multi-turn conversational flows to improve reliability and accuracy. Built RAG pipelines using Azure Cognitive Search, vector embeddings, chunking strategies, and metadata-based retrieval to deliver grounded, domain-specific answers for document-heavy enterprise use cases.

Statistical Software Development in R & Python: Developed production statistical computing solutions using R and Python supporting enterprise analytics serving 10K+ users; implemented data processing pipelines, statistical models, and machine learning algorithms in both languages; optimized computational performance through efficient data structures, vectorized operations, and algorithmic improvements reducing runtime by 40%; designed modular, reusable code libraries following software engineering best practices including version control (Git), unit testing (testthat in R, pytest in Python), and comprehensive documentation.

Data Structure Optimization & Package Development: Engineered advanced data structures in R and Python optimizing memory usage and performance for large-scale datasets (10M+ records); developed and maintained internal packages and libraries supporting analytical workflows; implemented efficient algorithms for data transformation, aggregation, and statistical computation; applied profiling tools (R profvis, Python cProfile) identifying performance bottlenecks and implementing optimization strategies; created scalable solutions supporting both interactive analysis and production deployments.

Data Engineering & Big Data Pipelines: Engineered large-scale data pipelines using Azure Data Factory and Azure Databricks; developed PySpark ETL jobs on Spark clusters, optimized distributed transformations, and implemented Delta Lake–based architectures on Data Lake Storage (ADLS) to ensure reliable, versioned, and ML-ready datasets.

Research Software Engineering & Repository Development: Built data repository systems managing large-scale datasets with versioning, metadata tracking, and quality validation; developed automated workflows for data ingestion, transformation, and analysis using R and Python; implemented parallel processing frameworks (R parallel, Python multiprocessing) for computationally intensive statistical operations; established software development best practices including code review, continuous integration, and automated testing; maintained production systems with comprehensive monitoring ensuring reliability and correctness of statistical computations. Created data visualizations using ggplot2 and matplotlib for technical and non-technical audiences. Applied statistical methods using R (tidyverse, caret, forecast) and Python (scikit-learn, statsmodels, SciPy) for hypothesis testing, regression modeling, time-series forecasting, and experimental design; developed reproducible statistical analysis workflows producing comprehensive reports and visualizations.

Data Pipeline Architecture & Azure Databricks Engineering: Engineered large-scale data pipelines using Azure Data Factory and Azure Databricks supporting enterprise AI applications; developed PySpark ETL jobs on Spark clusters processing 100K+ records daily, optimized distributed transformations with partitioning strategies, and implemented Delta Lake architectures on Azure Data Lake Storage (ADLS) ensuring reliable, versioned, analytics-ready datasets with 99.9% uptime across production environments.

Data Modeling & Database Optimization: Designed scalable data models supporting business intelligence and analytics workloads using Microsoft SQL Server and Azure cloud platforms; implemented indexing strategies and query optimization best practices reducing data retrieval times by 40%; collaborated cross-functionally with product managers, analysts, and engineers to translate business requirements into technical data architectures meeting governance and performance standards.

Document Intelligence & Cognitive Automation: Integrated Azure Document Intelligence, Cognitive Search, Blob Storage, and Function Apps to automate document ingestion, semantic indexing, and large-scale retrieval workflows. Implemented machine learning and computer vision models using TensorFlow, Keras, and OpenCV for classification.

Advanced Analytics Infrastructure & Semantic Search: Developed semantic search and retrieval systems using Azure Cognitive Search, vector embeddings, and metadata-based indexing to support analytics use cases; implemented chunking strategies and distributed data processing optimizing information retrieval performance; built scalable backend microservices using Python (FastAPI) and SQL supporting cross-database lookups, data aggregation, and structured content extraction for business intelligence applications.

Backend Engineering & Intelligent Retrieval: Developed scalable backend microservices supporting AI inference, vector search, prompt routing, and hybrid retrieval (keyword + embedding) for cross-database lookups, multi-document summarization, and structured content extraction using Python (FastAPI), Azure Cloud, and Microsoft SQL Server.

Production Systems Monitoring & Performance Enhancement: Deployed scalable data systems on Azure Cloud with comprehensive monitoring pipelines tracking data quality metrics, pipeline performance, and system reliability; implemented CI/CD workflows automating deployment processes; displayed strong ownership by iterating on data architectures based on performance analytics and stakeholder feedback; worked in fast-paced environment with regular releases, continuously enhancing data capabilities to solve business-critical problems.