Hey! I'm Srinidhi, currently doing my Master's in Applied Data Intelligence at San JosΓ© State University. I'm one of those people who gets genuinely excited about making things work better with code and data. Whether it's building software that scales, wrangling messy datasets into something useful, or training ML models that don't just look good on paper but actually solve problems, I'm in.
Lately, I've been kind of obsessed with Agentic AI workflows. You know, those systems where AI agents actually collaborate and get stuff done autonomously? I'm actively working on building and orchestrating these workflows, trying to push what's possible when you let intelligent agents work together. It's wild seeing them reason through problems and execute tasks that would normally need constant human oversight.
What I'm looking for: Summer 2026 internships where I can actually build cool stuff:
- Software Engineering | I want to write code that thousands (or millions) of people use
- Data Science | Turning messy data into clear insights is my jam
- Machine Learning Engineering | Building ML systems that work in production, not just notebooks
- AI Engineering | Especially anything involving agentic systems or next-gen AI applications
π Location: San Jose/SF Bay Area, CA | π§ Email: [email protected] | π± Phone: +1 (669) 243-7083
Here's some work I'm actually proud of. Check out the repos if anything catches your eye (and maybe drop a β if you vibe with it):
|
Okay so this was fun. I built a complete serverless pipeline that pulls YouTube metrics, converts everything from JSON to Parquet (huge performance win), and pushes it all to dashboards that product teams can actually use. Results: 90% less manual data prep, 5x faster queries with Parquet optimization, sub-second insights for 10k+ records Stack: AWS Lambda, Glue, Athena, S3, Apache Airflow, Spark, Tableau, QuickSight |
Built this RAG-powered assistant that searches through 10k+ old support tickets to answer new questions. Basically like having the entire support history at your fingertips, instantly. Customers were impressed with the 95% answer accuracy. Impact: 70% faster resolution times, sub-second responses with actual source citations Stack: FastAPI, Next.js, Sentence Transformers, ChromaDB, Google Gemini API |
ποΈ Azure Medallion Data PipelineThis one was about doing data engineering properly. Implemented a full Bronze-Silver-Gold lakehouse architecture on Azure Databricks. The kind of thing that actually scales when you're pulling in 100k+ records from 8 different sources. Performance: 60% faster processing, now powering analytics for over $2M in revenue decisions Stack: Azure Databricks, Data Factory, Kafka, Power BI, Synapse |
Ever need to find sensitive info in images or docs? Built a CNN + OCR system that does exactly that, super useful for data governance teams. Drag-and-drop web interface so no technical skills needed. Accuracy: 98% F1 score on 600 images, 4-layer CNN with ~6.7M parameters Stack: TensorFlow.js, Python, Tesseract OCR, Docker |
|
This is where things get interesting. I experimented with AI agents that actually collaborate. They designed, coded, and deployed a trading simulation together. Watching them work is honestly pretty cool, especially with 95% test coverage across 25+ edge cases. Efficiency: Cut the dev cycle by 60%, prevented 100% of invalid transactions Stack: CrewAI, Python, Gradio, OpenAI GPT-4 |
Built this for aviation safety. Predicts low-visibility events using weather data and physics-based features like dewpoint calculations. Added custom meteorological transformers because standard ones weren't cutting it. Performance: 45% better RMSE, 25% accuracy boost in critical low-visibility conditions Stack: scikit-learn, Flask, Docker, Python |
|
Built an agentic system that automates financial research. It scrapes web data, processes unstructured info, and generates structured investment reports. Reduced hours of manual work to just minutes. Speed: 95% reduction in research time, 90% report accuracy with 100% automated processing Stack: CrewAI, OpenAI GPT-4, Python, SerperDev API |
Analyzed 10k+ LLM performance logs to figure out which AI models work best for specific agent types. Built clustering algorithms to match the right LLM to the right task, saving both time and compute costs. Optimization: 40% performance improvement, 60% faster decision-making, 95% data quality Stack: NumPy, Pandas, Python, Clustering Algorithms |
|
Built a content-based recommendation engine analyzing 8,807 Netflix titles. Used TF-IDF vectorization and cosine similarity to suggest movies based on what you've watched. Pretty fun exploring the data patterns too. Scale: 8,807 titles analyzed, content-based filtering with genre and geographic pattern insights Stack: Python, scikit-learn, Pandas, matplotlib, seaborn |
Built and deployed an XGBoost churn model on AWS SageMaker that hit 84% precision. This actually helped cut customer attrition by 12% and bumped renewals by 15%. Seeing it work in production was rewarding.
Created serverless ETL pipelines with Lambda, Glue, and Airflow processing 25k+ records every single day. Improved data quality by 30%. Turns out clean pipelines equal clean data (who knew? π).
Spent time optimizing our Redshift warehouse by playing around with schemas and partitions until we got queries running 50% faster and slashed storage costs by 30%.
My first real taste of production data work. Built various ML models (both supervised and unsupervised) that beat our baseline by 18%. Learned a ton about what actually works vs. what just looks good in theory.
Engineered pipelines for multi-million-row datasets and got preprocessing time down by 35%. Turns out optimization matters when you're dealing with that much data. Also did comprehensive EDA which improved our feature engineering by 20%. Good old exploratory analysis never goes out of style.
This is what I work with day-to-day (and keep adding to). Always learning something new!
What drives me is pure curiosity, honestly. I can't help but tinker with new frameworks, read research papers at 2 AM (yes, really), and constantly ask "what if we did it differently?" I genuinely read research papers for fun. Yeah, I know how that sounds. But there's something satisfying about understanding how things work at a fundamental level.
I love the full spectrum, from writing clean production code to experimenting with bleeding-edge AI concepts. Here's a hill I'll die on: good data engineering is like 80% of what makes ML projects succeed. Clean pipelines beat fancy algorithms every time.
Currently exploring: Agentic AI workflows and multi-agent orchestration. Actively building systems where AI agents collaborate, reason, and execute complex tasks. It's fascinating watching them reason through complex problems. Like a really smart team that never sleeps. I really believe we're at the start of something huge with AI. Want to be someone who builds it, not just talks about it.
Also diving into advanced RAG architectures (going way beyond basic retrieval with hybrid search and re-ranking), real-time data streaming with Kafka and Flink, privacy-preserving ML techniques (because training models on sensitive data shouldn't mean exposing that data), system design at scale (learning how to build distributed systems that don't fall apart when things get big), and LLM fine-tuning to make AI models actually useful for specific tasks.
M.S. in Applied Data Intelligence | San JosΓ© State University | Jan 2025 - May 2027
Taking/took classes in: Data Warehousing & Pipelines, Big Data Tech, Machine Learning, Deep Learning, Statistical Analysis, Generative AI Applications, Data Mining, Data Viz
B.E. in Electrical & Electronics Engineering | PES College of Engineering, India
Kind of obsessed with Agentic AI right now. Building systems where AI agents work together autonomously feels like we're living in the future. If I have to do something manually more than twice, I'm writing a script for it. Or lately, building an agent to do it.
Always have 3-4 side experiments running. Currently it's multi-agent systems and advanced RAG architectures. This week I'm exploring prompt engineering techniques. Last week was Kafka optimization. Learn something new every day philosophy.
Guilty pleasure: spending way too long making the perfect data visualization. Color schemes matter, okay? Run on coffee and curiosity. Mostly curiosity about why things work the way they do. I like both sides of the coin. Building rock-solid production code AND experimenting with wild, experimental ideas. Why choose?
Still figuring out privacy-preserving ML but it's important work. Reading everything I can get my hands on. Research papers, tech blogs, docs, you name it. If it's about AI, data, or software engineering, I'm probably reading it.
Seriously, I love talking about this stuff. Whether you're hiring, building something cool, or just want to geek out about tech, hit me up!
"Data is the new oil. It's valuable, but if unrefined it cannot really be used." β Clive Humby
Thanks for checking out my profile! If you see a project you like, drop a β β makes my day every time π