Name: Manimaran K
Role: Systems & Quantum AI Engineer | Undergraduate Researcher
Degree: B.Tech in Computer Science & Business Systems (CSBS)
Institution: Rajalakshmi Institute of Technology, Chennai, India
Core_Domains:
- Hybrid Quantum-Classical Algorithms (Qiskit VQE, Molecular Hamiltonians)
- Multi-Agent Autonomous Systems & Financial Governance (Consensus, Audit Chains)
- Physics-Informed Deep Learning (PINNs, Atmospheric Reentry Dynamics)
- High-Throughput Distributed Systems (Solana CLMM, Rust)
- Scalable Full-Stack Engineering (FastAPI, Next.js, Docker, Microservices)
Mindset: "Engineering from foundational physical & quantum algorithms to scalable distributed architectures."
End-to-End In Silico Biophysical & Quantum Chemical Hit-to-Lead Screening Engine
- Generative Chemistry: 2-layer SMILES LSTM trained on ZINC chemical libraries for de novo synthetically accessible drug scaffold generation.
-
Biophysical Pose Optimization: RDKit
ETKDGv23D stereochemical embedding with MMFF94 force-field minimization and SciPy boundedL-BFGS-BLennard-Jones + Coulomb electrostatic docking against target PDB crystals. -
Quantum Ground-State Energy: Parameterized Qiskit 1.x VQE (
$CAS(4,4)$ active-space Hamiltonian mapped via Jordan-Wigner / Parity transformations) with real-time variational circuit simulation. -
Thermodynamic Rigor: Statistical thermodynamic binding affinity (
$\Delta G_{\text{binding}}$ ,$K_d$ ) coupled with a virtual in vitro Hill-Langmuir pharmacodynamic twin. -
Stack:
Python·Qiskit 1.x·PyTorch·RDKit·SciPy·TypeScript·Next.js·FastAPI·Docker·Railway·Render
[ZINC SMILES LSTM] ──> [RDKit MMFF94] ──> [L-BFGS-B Docking] ──> [Qiskit VQE Solver] ──> [Hill-Langmuir PD Twin]
Multi-Agent Financial Risk Governance with Real-Time Consensus & Cryptographic Audit Chains
- Quantitative LLM Council: Six specialized domain agents (Return, Liquidity, Cost, Volatility, Capital, Risk) executing quantitative checks and LLM-backed qualitative vetoes.
- Consensus & Circuit Breakers: Weighted dynamic majority quorum with strict drawdown breakers and unyielding mandate enforcement.
- Tamper-Proof Audit & Voice Telemetry: SHA-256 hash-chained immutable audit log; integrated with Stitch double-entry ledgers and ElevenLabs real-time voice constraint interfaces.
- Stack:
Python·FastAPI·Multi-Agent Architectures·PostgreSQL·Redis·Docker·ElevenLabs
High-Performance Concentrated Liquidity Market Maker Protocol on Solana
- Concentrated Liquidity Math: Custom tick math and dynamic range-bounded liquidity provisioning algorithms for capital-efficient DEX architectures.
- On-Chain Security: Program-derived address (PDA) vault segregation, slippage protection, and sub-second fee compounding execution.
- Stack:
Solana·Rust·Anchor Framework·DeFi Primitives·TypeScript
Physics-Informed Neural Networks for Spacecraft Aerodynamic Decay Modeling
- Physical Law Integration: Direct embedding of upper atmospheric drag, orbital mechanics, and gravitational ODEs into neural loss gradients.
- Generalization: Outperformed naive black-box regressors across sparse and irregular satellite telemetry regimes.
- Stack:
Python·PyTorch·PINNs·SciPy·Orbital Mechanics
Deep Learning Sequence Modeling for High-Volatility Financial Assets
- Built an LSTM recurrent neural network architecture designed to resolve complex long-range temporal dependencies in equity pricing.
- Automated pipeline for feature scaling, sequence sliding-window formulation, and multi-horizon forecasting.
- Stack:
Python·TensorFlow/Keras·LSTM·Pandas·NumPy
- Quantum-Classical Hybrid Algorithms: Advancing VQE active-space formulations for complex biomolecular binding pocket calculations.
- Scientific Publications: Authoring peer-reviewed works bridging Physics-Informed Neural Networks (PINNs) and quantum chemistry.
- Autonomous Governance & Multi-Agent Consensus: Advancing verifiable LLM council architectures and cryptographic audit chains.
- Ultra-Low Latency Systems: Expanding high-frequency decentralized financial primitives on Solana.
