Sai Krishna Ponnam👋 Hello, I'm
🚀 Master's in Computer Science @ UW-Madison
🌟 Passionate about Machine Learning and building intelligent and scalable applications.
About Me
I am a Master's student in Computer Science at the University of Wisconsin-Madison, specializing in natural language processing, machine learning, and distributed systems. Before grad school, I spent 5 years at Target Corporation as a software engineer, where I worked on large-scale search systems, real-time data pipelines, and low-latency production services used by millions of users.
My work spans building low-latency REST APIs using Kotlin, Micronaut, Python and Kubernetes; designing real-time and offline data pipelines with Kafka and Spark; and fine-tuning LLMs like LLaMA 2 for Named Entity Recognition using PyTorch and QLoRA. Beyond my professional work, I actively lead learning groups, mentor aspiring developers, and enjoy exploring emerging ideas and reading cutting-edge research papers and building practical systems inspired by them.
When I am not immersed in code or coursework, you will probably find me playing badminton or swimming, relaxing with manga, or writing about the ML papers and concepts I explore in my notes. I am always curious, always learning, and passionate about sharing what I learn with others.
Key Skills Summary
Experience
Senior Software Engineer
Target Corporation, India
Nov 2021 - Jul 2024
Part of the Guided Search team focused on enhancing shopper experience through intelligent query understanding and dynamic search guidance, including facets, autocomplete, and related suggestions.
Led development of a lexical-based NER system in Python to extract query attributes, reducing manual effort by 50%. Later boosted performance by fine-tuning LLaMA2-70B with QLoRA, improving attribute coverage by 12%.
Built a facet interaction metrics pipeline using Scala Spark, reducing analysis time by 80% and enabling data-driven roadmap planning.
Developed a A/B testing analysis framework in PySpark, streamlining experimentation across features and cutting analysis time by 40%.
Software Engineer
Target Corporation, India
Jul 2019 - Oct 2021
Member of the SearchBox team, focused on delivering high-performance and context-aware autocomplete suggestions to improve user engagement and search efficiency.
Implemented a low-latency Python service for LSTM-based context model for autocomplete, increasing demand by 2% (~$30M/year).
Automated deployment of a multi-node Redis cluster on Kubernetes, cutting deployment time from 1 hour to 10 minutes.
Optimized search services using JMeter, reducing monthly operational costs by 15%.
Projects
A collection of my notes on machine learning concepts and summaries of papers I've read, hosted on GitHub Pages.
Developed a neural network framework in C++ with CUDA, using shared memory and reduction techniques for optimization. Implemented core operations like matrix multiplication, activation functions, and backpropagation.
Implemented a Vocabulary-Free Multilingual Neural Tokenizer. Utilized FVT and FOCUS to adapt a large tokenizer into a language-specific one for improved factual knowledge retrieval.
Technical Skills
Kotlin
95%Python
95%Java
85%C++
80%Scala
90%SQL
80%PyTorch
90%Scikit-learn
85%Pandas
90%NumPy
90%HuggingFace
75%CUDA
50%Elasticsearch
90%Lucene
90%Solr
85%Docker
80%Git & GitHub
90%Kubernetes
75%Spark
90%Kafka
90%Bash
90%Get In Touch
I'm always open to discussing new projects, creative ideas, or opportunities to be part of an amazing team. If you have a question, want to say hi, or would like to collaborate, I'd love to hear from you.
ponnam2@wisc.edu