Curriculum Vitae

Cross-functional team of 30+ students developing India's first Level 5 self-driving car, customized for Indian roads.

• F1TENTH Autonomous Racing: Engineering a LibTorch-Acados C++ runtime to deploy learned-dynamics HyperMPC at a 100 Hz control loop on a 1/10-scale racecar.
• Adaptive Parameter Learning: Developing a GRU-based Hyper-Param network with B-spline control encoding to learn horizon-varying tire and kinematic model parameters.
• Autonomous Golf Cart Localization: Building an autonomous localization stack for reliable outdoor navigation using high-precision RTK GNSS.
• Joint Sensor Fusion (EKF): Developing a Joint Extended Kalman Filter fusing RTK GNSS & IMU measurements to estimate vehicle 6-DOF pose, linear velocity & heading in real time.

Selected as 1 of the 12 WQBRAIN Research Consultants (Rising Stars cohort) globally to attend the IQC'25 Finals in Singapore.

• Master Consultant (Q3'25): Achieved Master Consultant status via a portfolio of orthogonal signals yielding an after-cost out-sample Information Ratio (IR) of 2.39.
• 320+ Submitted Alphas: Maintained an average self-correlation of 0.25 with out-sample fitness scores up to 6.47.
• Composite Strategy Architecture: Engineered 90+ composite alpha strategies by selecting, filtering & combining orthogonal quantitative signals.
• Automated Research Pipeline: Automated alpha factor research via an LLM-driven iterative refinement pipeline for large-scale systematic signal generation.

• Built a regime-switching portfolio optimizer integrating Gaussian HMMs, Random Forest meta-models & convex allocation across 503 equities.
• Constructed event-driven trading labels using CUSUM sampling, Triple Barrier labeling & volatility-normalized dynamic thresholds.
• Engineered 250+ alpha features spanning price, volatility, momentum, volume & macroeconomic factor spaces.
• Mitigated temporal leakage using Purged 5-Fold Cross-Validation with 1% embargoes & sequential bootstrapping.

• Created a multi-agent AI system in a 7-member team for autonomous sky exploration, transient detection & hypothesis generation.
• Built an astronomy benchmark pipeline integrating Google Gemini with academic literature retrieval & fault-tolerant LLM orchestration.
• Constructed a semantic Neo4j knowledge graph using transformer-based claim de-duplication with 95% similarity threshold.
• Engineered a high-throughput Swift FITS pipeline extracting temporal, spectral & physical features from astronomical observations.

• Built an EDSR with 32 residual blocks, boosting training throughput via mixed-precision (FP16/FP32) using PyTorch AMP.
• Trained over 290k iterations using L1 loss, achieving 34.84 dB PSNR & 0.939 SSIM on DIV2K.
• Bounded peak VRAM spikes via spatial tiling & AMP, deploying sub-pixel convolution for fast upsampling; benchmarked SRCNN on Sen2Venus (37.47 dB PSNR).

• Developed a Direct Hybrid Collocation PINN estimating Young's modulus fields across 3 Additive Manufacturing lattice specimens.
• Unified 8 DIC anchors with 10,000+ Sobol collocation points, strictly enforcing 2D Cauchy Momentum PDE equilibrium constraints.
• Solved the ill-posed inverse elasticity problem via a 7-objective composite loss reconstructing displacement, stress, strain & modulus fields.

• Engineered a mid-frequency Bitcoin options simulator with slippage-aware logic and Delta Exchange API integration.
• Built an event-driven execution engine for asynchronous ccxt API interactions, minimizing execution overhead.
• Designed dynamic strike selection & trade lifecycle management with automated PnL, turnover, slippage & Greek exposures.
• Deployed Multi-Armed Bandits (UCB / Thompson Sampling) for hyperparameter tuning, balancing exploration and exploitation.

• Developed a hybrid Neural-MPC controller for a 6-DOF active suspension using OSQP, Joint EKF & residual learning.
• Distilled a 40-step MPC into a 4,736-parameter neural policy using 20k expert trajectories, reducing peak angular velocities up to 3x.

• Built virtual sensing framework for Spatial ANC using signals from 4 remote microphones for pressure reconstruction.
• Estimated virtual microphone pressure by training an ObsTasNet observation-filter on 400 simulated acoustic scenes (12.6x faster than kernel methods).

• Implemented the Black-Scholes model for European options using stochastic processes and risk-neutral valuation in Python.
• Performed sensitivity analysis on volatility, interest rates & maturity (Greeks); explored martingale pricing & finite-difference methods.

• Responsible for effectively managing an annual operating budget of INR 4+ Million, allocating resources for optimal outcomes.
• Coordinating communication between Technical Core & Non-Technical teams; spearheaded a 2-month recruitment drive for 100+ aspirants.

• Guided 9 students with suitable references & materials to build theoretical understanding of Algorithmic Trading and Derivative Pricing.

• Mentored 20+ students in building algo-trading protocols & validating system dynamics within quantitative trading engines.