Full Stack AI Engineer at BitFicial

  • Built a Voice AI agent using VAPI, Make.com, and Zapier, capable of leveraging multiple agent tools to perform a wide range of tasks — including sending SMS updates, booking appointments, processing user inputs, retrieving data based on queries, and responding with actionable insights, among other capabilities.
  • Engineered a user-specific AI assistant using OpenAI (GPT-4), LangChain, and a suite of Python data processing libraries (pandas, NumPy, scikit-learn) within a Django backend. Integrated with Apple Health data pipelines to ingest, normalize, and analyze multi-dimensional health metrics (sleep, heart rate, HRV, steps, etc.) in real time. Designed and implemented a LangChain-powered agent architecture with dynamic tool-calling capabilities using custom tools and toolkits. Built contextual memory modules and vector-based retrieval systems for user-specific data embeddings (via Pinecone), enabling natural language query resolution on personal health insights.
  • Implemented robust backend infrastructure for:
    • Stateful chat and session context management
    • Token optimization strategies to stay within LLM context window limits
    • Access control layers ensuring strict user-level data isolation
    • Tool routing logic to invoke the appropriate function/tool based on user intent
  • This solution enabled true chat-with-your-own-data capabilities: the LLM agent could only access and reason over data tied to the authenticated user (metrics, historical chats, preferences, and behavioral patterns), resulting in a hyper-personalized virtual health assistant experience.
  • Developed an all-in-one, tool-calling agent using Python and LangChain, designed to orchestrate multiple tools and workflows within a unified agent architecture. Key capabilities included:
    • User-specific query resolution: Dynamically queried relational databases to retrieve and respond with user-specific data across multiple tables and schemas.
    • Transactional data lookup: Implemented structured retrieval mechanisms to fetch matching records from high-volume transactional databases using optimized SQL queries.
    • Modular API integration: Integrated multiple third-party and internal replier APIs, with tool-based routing logic to handle diverse use cases (each use case mapped to a distinct tool).
    • Research & analysis pipeline: Built tools for contextual topic research, aggregating data from APIs, knowledge bases, and LLMs to generate comprehensive insights or feed downstream processes.
    • Automated reporting engine: Designed and implemented a reporting toolchain where metadata for various reports is generated via LangChain tools and transmitted to an endpoint for PDF generation using WeasyPrint (HTML to PDF conversion).
  • This agent served as a scalable and extensible backbone for automated data-driven workflows, API orchestration, and personalized user interactions.

Senior Data Engineer at BitFicial

  • Built a sophisticated royalty dashboard application from scratch. Leveraging Django as the backend framework and Django's templates for the frontend, I seamlessly integrated various components to create a seamless user experience. The project required handling large volumes of data, necessitating efficient data management techniques. To automate recurring tasks, I implemented cron jobs, ensuring timely updates and data synchronization. Additionally, I utilized JavaScript and Python to enhance the application's functionality and interactivity. Deploying the application on AWS further demonstrated my proficiency in cloud computing and scalability. This project not only challenged me to think critically and creatively, but it also fortified my skills in full-stack development and data handling.
  • Managed data team on a groundbreaking project that entailed the deployment of a machine learning model API and a robust backend. Python served as the primary language for developing the machine learning model API, which was meticulously crafted to deliver accurate predictions and insights. Leveraging my expertise in cloud computing, I successfully deployed both the model API and the backend on the Google Cloud Platform (GCP), ensuring scalability and high availability. To streamline the data flow and processing, I implemented efficient data pipelines on GCP, facilitating seamless integration between the model API and the backend. This project challenged me to apply my skills in Python programming, ML, and cloud deployments.
  • Worked on an impactful project that involved data synchronization from various sources using SyncHub. I effectively integrated data from platforms such as Xero and Unleashed, ensuring accurate and up-to-date information for analysis. Leveraging my expertise in data manipulation and analysis, I generated valuable insights from the integrated data sets. To effectively communicate these insights to clients, I leveraged Power BI to create visually appealing and informative dashboards. These dashboards provided clients with a comprehensive view of their business performance and key metrics, empowering them to make data-driven decisions. This project allowed me to demonstrate my proficiency in data analysis, data integration, and visualization, showcasing my ability to deliver actionable insights to clients.

Python/Data Engineer at Developers Studio

  • Leading data engineering team, I oversaw the collection of diverse data types from multiple sources and unified them into a single database. We utilized advanced data integration techniques, including ML algorithms, to merge, cleanse, and standardize the data. Our efforts resulted in a user-friendly search engine that met our specific requirements and provided reliable and accurate information.
  • Optimized the cron jobs and reduced execution times by 30%, enabling the data team to run the jobs every hour instead of once a day. This improved data freshness and availability. I analyzed the existing jobs, streamlined the process, and eliminated unnecessary steps to create a more efficient workflow.

Data Engineer at ML Sense

  • Worked with the data science team, and wrote multiple Python/PySpark scripts, to fetch data from different apis, and scheduled them to run daily and append the data into Azure Databricks Spark tables.
  • Converted raw JSON data streams, coming into the data lake, into analyzed dashboard in Azure Databricks. Cleaning was done by scheduled Python scripts and SQL queries.
  • Created GCP Cloud Audit tool with the help of Cloud Audit logs, BigQuery and Cloud Functions. The tool provides complete and precise information about each resource, created in a project. The information contained Resource Name, Creation Time, Update Time, Creator (Service Account/User), Project, Cost, Data Processed etc.
  • Created an ETL pipeline using Cloud Functions, Scheduler and Bigquery. This procedure gradually scraped almost 27 million records from the source, cleaned incoming data, and then stored it in Cloud SQL on daily basis. This data was then used for a dashboard for a client.
  • After in-depth data analysis, designed and created interactive dashboards for multiple clients in Tableau.
  • Used Django and React to create a web application, deployed to GCP. Scheduled data fetch from Facebook API, Trueclicks and Google Display and Video (DV 360) and wrote python scripts in Django backend and Cloud Functions to analyze the data and display the information on frontend using React charts.
  • Successfully migrated complete infrastructure, code and data of an organization from one GCP project to another. Created pipelines for the data to be synchronized across projects, so that data can be shared between them.
  • Debugged and resolved multiple issues in multiple GCP projects for both Production and Development environments. Debugged cloud services including Terraform deployments, Cloud Composer, Airflow, Cloud Dataflow, Cloud Functions etc.
  • Using Terraform, Github Actions and GCP (Composer Airflow), created end-to-end Data Pipeline, reading, processing and storing almost 7 million records each month. Terraform scripts get triggered by GH Actions, and create the Infrastructure i.e., Composer Environment, and then GH Actions update Airflow DAGs.
  • Used Terraform to create Datadog monitors for a DAG, running in Airflow.
  • Migrated a data pipeline that was previously running on GCP's Dataflow in Scala to a Python-based pipeline deployed on a GCE instance, saving approximately 6000 GBP per annum. The migration required a deep understanding of both Scala and Python, as well as expertise in GCP's infrastructure. We optimized the pipeline's performance to handle the same workload as the previous pipeline, without sacrificing speed or accuracy, resulting in significant cost savings for our organization.

Python Engineer Intern at ML Sense

  • Developed a Django React app and created complete CICD pipeline using GitHub Actions to deploy on Google App Engine.
  • Created multiple Python scrapers using Selenium and BS4, and completed a GUI based application, able to handle all the scrapers from a single User Interface.

Freelancer at Fiverr

  • Worked on various client projects including Scraping, Web development, Machine Learning and Database Designing.