Hey there! I’m Karthik, an MS CSE graduate from Georgia Tech with seasoned proficiency in adapting ML paradigms to chip design at Xilinx and AMD. Over a span of 3 years at Xilinx and AMD, I lead ML R&D at the Timing Closure team, contributed multiple features to Vivado™, and published patents in continual learning of FPGA compile-time estimators, compute resource optimization in ASIC emulation, and FPGA placement parameter ranking to aid timing closure. Most recently at Georgia Tech, my work on “incentivizing neuro-symbolic reasoning in vision-language foundation models” engendered by AlphaGeometry, DeepSeek-R1, etc. has aided the fortification of my perspicacious aptitude for literature review and effectuation of SoTA compute-constrained post-training techniques in MLLMs.
I’m actively seeking job opportunities in applied ML (across roles in search, recommendations, RAG, LLM fine-tuning, LLM eval, data generation, post-training RL, reasoning, tool-use, etc.). Please feel at liberty to reach out to confer further!
projects
Estimating fanout wirelength in FPGA designs via Sequence Modeling
AI Intern @ AMD (San Jose) | Summer '25
GNN-RL methodology to optimize routing sequence in custom GPU designs
GRA @ GTCAD Lab, Georgia Tech
publications
-
Game-based digital intervention to aid cognitive development in autism spectrum disorderNature Sci. Rep., 2021, associated with Lab of Neural Systems, IIT Kanpur
hobbies
Breathing, blinking, slurping water, typing random keys on keyboard during bootup, listening to Lex Fridman podcasts on 3x, etc.