Suvanth Erranki

Built a competitive programming online judge supporting C/C++/Java/Python/PyPy with sandboxed execution, custom checkers, testcase generators, and automated scoring. Implemented isolated Linux-based grading environments with enforced CPU, memory, wall-clock, and process limits, evaluating 5K+ submissions from 500+ contestants across 3 competitions.

Developed a C++ chess engine using Monte Carlo Tree Search (MCTS) and bitboard-based move generation. Optimized move generation with precomputed attack tables, achieving 10M nodes/sec on a single Apple M2 Pro core while tuning PUCT hyperparameters.

Developed heuristic pathfinding and multi-agent coordination under strict bytecode budgets to optimize decision quality in hard real-time constraints.