Sravan Jayati

Software Developer · AI researcher · MS EECS, UC Merced

Thanks for choosing to know more about me!

I'm passionate about advancing AI to perceive our world in ways that can foster a better, more convenient society for humanity. In my career, I strive to explore open research problems in Computer Vision such as segmentation, generative models, multimodality, etc.

My current research at UC Merced involves vision-language models in the domain of building damage assessment post disasters like hurricanes and earthquakes. As part of my coursework as an EECS MS student, I worked on a variety of projects related to Signal Processing, Human Computer Interaction, Deep Learning and Reinforcement Learning.

As a Computer Vision and Machine Learning engineer prior to my Master's, I worked on developing and optimizing models for hydroponic farms and driver assistant systems. I also prototyped and maintained code processes and pipelines to enable these systems.

In my free time, I enjoy producing music and geeking out on thoughtfully designed products.

News

October 2024

  • Participated (as a team of 3) in the Ted AI 2024 hackathon held at Microsoft Reactor, San Francisco, where we won a side quest prize from Infinity AI for the best use of their tool.
  • Attended SIGSPATIAL '24 in Atlanta, GA (my first in-person conference!). Connected with researchers from diverse backgrounds on topics spanning geospatial technology, large language models, and computer vision.

September 2024

  • Submitted a paper for the GeoAI '24 workshop, part of the ACM SIGSPATIAL '24 conference - and it got accepted!

July 2024

  • Mentored a group of multidisciplinary undergraduates on a research project during a two-week program, the Data Science Challenge, at Lawrence Livermore National Lab.

May 2024

  • Officially started as a Graduate Student Researcher at the Computer Vision Lab, UC Merced.

February 2024

  • Began work on a building damage assessment project, in collaboration with Prof. Henry Burton, a civil engineering professor from UCLA.

October 2023

  • Presented my bachelor thesis work remotely for ICCMEH 2023.

September 2023

  • Joined the Computer Vision Lab under Prof. Shawn Newsam.

August 2023

  • Began my MS in EECS at UC Merced!

Experience

Graduate Student Researcher

UC Merced

Multimodal AI research for post-disaster building damage assessment as part of my master’s thesis.

  • Collaborated with UCLA’s Civil Engineering team to curate a dataset of hurricane-affected building images.
  • Published a workshop paper exploring ChatGPT-generated image descriptions combined with images to enhance vision-language models (e.g., ViLT, CLIP) compared to image-only models like ViT.
  • Developing a transformer-based pipeline integrating variable inputs, including images, text, and tabular data, to improve damage assessment accuracy.
Aug 2023 – Present

Team Lead, Data Science Challenge '24

Lawrence Livermore National Lab

Selected as a graduate team lead for a 2-week data science program, mentoring a team of undergraduates on ECG data analysis.

  • Designed and implemented machine learning and deep learning models for cardiac health pattern recognition.
  • Engaged in discussions with LLNL researchers on the future of AI and staying in AI research.
  • Presented findings via a poster session to LLNL researchers during the program's conclusion.
LLNL
Jul 2024

Computer Vision Engineer

AgEye Technologies

Developed computer vision and image processing pipelines for a rail-mounted sensor system scanning hydroponic crops.

  • Built plant segmentation pipelines using dynamic thresholding and watershed algorithms.
  • Implemented a YOLOv4-based real-time plant detection model for edge deployment on NVIDIA Jetson Nano.
  • Created a custom leaf instance segmentation network using Holistically-Nested Edge Detection and morphological operations.
  • Deployed vision pipelines to production and initiated weekly AI knowledge-sharing sessions.
AgEye
Jun 2022 – Jan 2023

Machine Learning Engineer

Neva Innovation Labs

Conducted R&D in machine learning and computer vision for driver behavior profiling and road condition analysis.

  • Developed a multi-branch Fully Convolutional Network for lane marking detection, leveraging DBSCAN clustering.
  • Built a road condition classification system using Fully Convolutional Networks, Gabor edge detection, and Support Vector Machines.
  • Implemented a windshield rain detection model using blob detection, optical flow, and Gradient Boosting.
  • Profiled driver behavior using IMU sensor data for fleet management and accident analysis in insurance.
Neva
Jan 2021 – May 2022

Education

University of California, Merced

M.S. - Electrical Engineering and Computer Science
  • Relevant courses taken: Signal Processing, Image Processing, Deep Learning, Reinforcement Learning, Compilers, Human Computer Interaction
  • Research at Computer Vision Lab
Expected 2025

Manipal Institute of Technology, Manipal, Karnataka

B. Tech. - Electronics And Communication
  • Thesis: Pen-based Handwritten Character Recognition for Kannada Numbers
  • Actively took part in various programming workshops
  • Played keys for my band, which qualified for an 'unplugged' event at a cultural fest
  • One of the founding members for an acapella group
May 2018

See Projects