Data Science · Analytics · Engineering

I'm Sathwick Kiran.I build data and AI systems that drive business decisions.

Data professional exploring AI, ML, and big data solutions, building pipelines, models, and dashboards that help teams move faster with better information.

Authorized to work in the US  ·  Open to relocation
Explore My Work Get In Touch
About

About Me

Sathwick Kiran

Hey! I am a data professional with 2 years of experience at Sony and an MS in Data Science from the University at Buffalo. My work spans data engineering, analytics, and applied machine learning. I build ETL pipelines, Power BI dashboards, and reporting solutions that cross-functional teams use to make decisions.

At Sony I worked on Python and SQL data workflows across 5+ internal teams, automated reporting pipelines, and put in place data quality checks that improved consistency and reduced manual effort by 15%. I also worked with AWS-based workflows and CI/CD automation, focusing on building reliable data infrastructure that improved reporting efficiency and stakeholder decision-making.

My academic and project work blends data engineering with advanced analytics, from building a LangGraph multi-agent AI system for financial market analysis to processing 3.3M+ records using PySpark and NLP on Hadoop, and building a containerized DataOps observability platform with Kafka, Airflow, dbt, and Tableau.

Outside of data and AI, I'm a drummer, a big cricket fan, and slightly obsessed with Formula 1. You'll usually find me watching qualifying sessions on weekends.

Capabilities

Skills & Certifications

Programming

Python SQL PySpark R Scala PyTorch

Data Science & ML

Machine Learning NLP Statistical Analysis Feature Engineering Scikit-learn Pandas EDA PyTorch

BI & Visualization

Power BI Tableau Excel Apache Superset

Data Engineering & Cloud

ETL / ELT Data Pipelines dbt Airflow Kafka Hadoop Apache Spark AWS Azure Databricks Docker Snowflake

Tools & Platforms

GitLab GitHub Jira CI/CD PostgreSQL LangGraph FastAPI Generative AI Prompt Engineering

Certifications

Azure Machine Learning for Data Scientist
Microsoft Azure
2026
Azure ML: Model Training & Evaluation
Microsoft Azure
2026
Databricks Fundamentals
Databricks
2025
Snowflake Data Warehousing
Snowflake
2025
AWS Cloud Practitioner
Amazon Web Services
2023
Career

Experience

Oct 2026 to Present

AI Engineer, Delivery Lead

CampusMesh
  • Leading a team of 2 with the CTO to deliver CampusIQ, an LLM discovery assistant on a platform of ~2K active students, while owning its full backend.
  • Rebuilt CampusIQ's LLM harness with a deterministic answer path and config driven Bedrock client, holding p95 latency under 3s and retrieval under 300ms.
  • Designed a 100-question evaluation set and per query audit records to test whether CampusIQ answers only from CampusMesh data, guiding prompt fixes.
  • Secured CampusIQ's AWS Lambda API, closing an open token endpoint and adding ownership checks on 5 handlers so students see only their own chats and files.

Technologies & Tools

Python AWS Bedrock AWS Lambda LLMs Retrieval LLM Evaluation API Security
Sep 2026 to Present

Data Scientist

Redprint · Amherst, NY
  • Developed an LLM evaluation harness for a gym aware workout assistant, scoring 300+ plans against live inventories and cutting equipment hallucinations 42%.
  • Engineered structured extraction over 4K+ workout histories and 300+ exercises, creating the feature layer behind personalized workout recommendations.
  • Built Python and SQL pipelines unifying NFC tap telemetry, workout logs, and equipment inventories across 20 gyms, matching 40K+ monthly taps at 97%.

Technologies & Tools

Python SQL LLM Evaluation Feature Engineering Data Pipelines NFC Telemetry
Feb 2026 to Aug 2026

Data Science Intern

Youro, LLC · Remote
  • Trained scikit-learn models on patient symptom features to predict severity scores and support specialist routing, achieving inference latency under 100ms.
  • Automated preprocessing and feature engineering pipelines with Pandas and NumPy, lifting training data quality 26% and saving 30% of preparation time.
  • Presented 5+ product KPIs from Tableau to the CEO, informing decisions to build a provider platform and keep data internal through RAG.

Technologies & Tools

Python Scikit-learn Pandas NumPy Tableau RAG
Jul 2025 to Dec 2025

Student Associate

University at Buffalo-SUNY

Buffalo, NY

Aug 2022 to Apr 2024

Software Engineer

Sony India Software Centre · Bengaluru
  • Performed exploratory data analysis on datasets across teams using Python and SQL to identify trends and support product and operations planning.
  • Engineered batch ETL pipelines in Python and SQL on AWS (Lambda, Glue, S3) across 5+ internal data feeds, saving 15% of manual reporting effort.
  • Developed reporting datasets and dashboard views for product, finance, and engineering teams, shortening request to insight turnaround by 20%.
  • Integrated GitLab CI/CD pipelines with Jira APIs to automate issue tracking and workflow updates, reducing coordination effort across teams by 20%.
  • Created a GitLab API dashboard for Sony's First Challenge to automate Wiki access requests and manager approvals, trimming manual coordination by 30%.

Technologies & Tools

Python SQL AWS (Lambda, Glue, S3) EDA GitLab CI/CD Power BI Tableau ETL Pipelines Jira
Feb 2022 to May 2022

Software Intern

LMARKS · Bengaluru
  • Collected, cleaned, and organized 500+ service and customer records using Excel and Python to support weekly reporting.
  • Prepared reports across 2,000+ customer entries to track service requests, job status, and recurring issue categories.
  • Analyzed 3+ years of historical installation, service, and customer data to identify follow-up gaps and support CRM process improvements for a business serving 2,000+ customers.

Technologies & Tools

Python Excel SQL Reporting CRM Git
Portfolio

Projects

DataOps Change Impact & Pipeline Failure Observability Platform

Architected and deployed an end-to-end DataOps observability platform on GCP using Docker Compose, engineering real-time pipeline monitoring from Kafka streaming through Airflow orchestration and dbt transformations across 2,185+ events and 7 data models. Built a RAG-based Failure Intelligence engine using FastAPI and Google Gemini AI to answer natural language queries on pipeline health, SLA breaches, and schema change blast radius, deployed with a live Chat UI on GCP. Designed 3 Grafana dashboards backed by dbt mart models with an Airflow DAG orchestrating automated breach detection, schema change impact scoring, and pipeline summary logging.

Python Kafka Airflow dbt PostgreSQL FastAPI Grafana GCP Gemini AI
Code

RideFlow Analytics: NYC Taxi Demand Forecasting with MLOps Pipeline

Built a LightGBM time series model on 3.5M+ NYC taxi trips forecasting hourly demand by location at MAE 2.895 and MAPE 9.23% for 50+ daily users. Designed a normalized AWS RDS schema with optimized SQL indexes, lowering query latency by 25% for an hourly refreshed Streamlit forecast dashboard. Streamlined AWS Glue and Athena workflows processing 3.5M+ records, reducing pipeline overhead by 40% and keeping data fresh for model training.

Python LightGBM AWS Lambda AWS Glue Athena AWS RDS Streamlit

Sentiment Wars: ML & Transformer Benchmarking for Sentiment Classification

Benchmarked TF-IDF machine learning models against fine-tuned DistilBERT and DeBERTa v3 on 10K+ Amazon reviews in a team of 3 to select the best classifier. Fine-tuned DeBERTa v3 base to 93.5% accuracy on Amazon review sentiment, outperforming DistilBERT at 91.8% and XGBoost at 88.3%. Logged 30+ model runs in MLflow to centralize evaluation metrics and reproducibility, speeding model comparison cycles by 30%.

Python Hugging Face Transformers DeBERTa DistilBERT Scikit-learn XGBoost MLflow Streamlit

AI Multi-Agent System for Market Intelligence (Synapse Street)

Led a team of 4 to build a LangGraph multi-agent system on 5GB of stock market data, flagging short-selling opportunities and generating readable investment summaries end-to-end in a 24-hour UB hackathon. Developed a Scikit-learn pipeline on OHLCV features with Qdrant vector search and sentence transformer embeddings, retrieving contextually similar market signals at Precision@10 of 0.60. Designed Streamlit and Tableau dashboards to surface model scores, agent findings, and backtest results for non-technical stakeholders.

LangGraph Python Qdrant Scikit-learn Hadoop HDFS Streamlit Tableau Vultr Cloud
Code Dashboard

Scalable Big Data Analytics using Hadoop, Spark and NLP

Designed and implemented a scalable pipeline for processing and analyzing 3.3M+ Amazon book reviews. Leveraged Hadoop (MapReduce, HDFS) and Apache Spark / PySpark for distributed data processing. Applied TF-IDF vectorization and NLP techniques for sentiment classification, rating prediction, and helpfulness scoring, achieving 95% accuracy post hyperparameter tuning via PySpark MLlib.

PySpark Hadoop Apache Spark NLP / TF-IDF Scikit-learn Python Docker
Code

Osteoporosis Risk Prediction using Machine Learning

Built a predictive modeling solution using demographic, lifestyle, and health data to assess osteoporosis risk and support clinical insights. Prepared 1,958 patient records, engineered features, and trained Logistic Regression, Random Forest, Decision Tree, and SVC models. Evaluated 4 models using accuracy, precision, recall, and F1 score, with Decision Tree achieving 90% accuracy and surfacing key risk factors.

Python Scikit-learn Pandas NumPy Random Forest Decision Tree
Code

Cryptocurrency Price Forecasting and Analysis using Machine Learning

Developed a machine learning pipeline to predict and analyze Bitcoin price fluctuations using large-scale financial datasets. The workflow involved extensive statistical analysis, feature engineering, and model experimentation to capture temporal price patterns. Containerized using Docker and deployed on AWS for scalability.

Python Scikit-learn Pandas Matplotlib Docker AWS
Code

Online Music Store Database System: Design, Normalization, and Query Optimization

Developed a relational database system for an online music store to manage sales, customers, and inventory efficiently. Designed and implemented an Entity Relationship (E/R) model and created 16 normalized tables (post-BCNF decomposition) with indexing strategies for performance optimization.

SQL PostgreSQL E/R Design Indexing BCNF
Code

Stock Market Analysis and Prediction using Machine Learning (NVDA Case Study)

Worked with NVIDIA (NVDA) stock price data to perform data cleaning, feature engineering, visualization, and predictive modeling. Applied KMeans clustering and developed predictive models using Linear Regression, Ridge Regression, and SVM, achieving R² scores of 1.000 and 0.997.

Python Pandas Scikit-learn Matplotlib KMeans SVM
Code

BMSCE_ONE: Centralized University Platform with ML Chatbot

Built a centralized digital platform to unify all university activities into a single hub, enabling students and faculty to seamlessly access announcements, academic schedules, and services. Implemented a machine learning based chatbot to handle FAQs and provide instant support.

Flutter Python Flask MongoDB ML Chatbot Android
Code

Indoor Navigation Using Augmented Reality

Developed an augmented reality based indoor navigation application to help students and visitors navigate campus buildings. Leveraged Unity with C# and LiDAR technology to build precise 3D maps, implementing Dijkstra's algorithm for accurate navigation. Nominated for Best Final Project award.

C# Unity Dijkstra's Algorithm AWS AR LiDAR
Private Repository
Academics

Education

Aug 2024 to Dec 2025

Master of Science in Data Science

University at Buffalo, State University of New York

Specialized in Advanced Machine Learning, Advanced Data Analytics, Statistical Analysis, and Big Data Applications.

Coursework & Tools

Python Machine Learning R Statistical Analysis Big Data Deep Learning LLM Data Intensive Computing MLOps SQL Data Visualization
Aug 2018 to May 2022

Bachelor of Technology, Computer Science

B.M.S. College of Engineering · Bengaluru

Specialized in full stack development, Operating Systems, Advanced Data Structures and Algorithms.

Technologies & Tools

Full Stack Data Structures Algorithms Augmented Reality Unity Python C++ Java C#
Contact

Get In Touch

Open to full-time roles in Data Analytics, Data Engineering, Analytics Engineering, and Data Science. Let's connect.

sathwickkiran04@gmail.com
+1 (716) 359-1432
New York, United States
Authorized to work in the US  ·  Open to relocation