SS

Projects

Engineering case studies — each project is documented with the problem, architecture, challenges, benchmarks, and lessons learned.

Distributed ML Training Framework

Built a distributed training framework that orchestrates heterogeneous consumer laptops over a LAN, enabling synchronized PyTorch Distributed Data Parallel (DDP) training with automated node management, scheduling, and telemetry.

PyTorch DDPPythongRPCSQLiteDockerAWSNCCLGloo
View Case Study
SYSTEM ARCHITECTURECluster management + PyTorch DDP worker planejob submissionstategRPCgRPClaunchlaunchtelemetrytelemetryUser / CLISubmit training jobCluster ManagergRPC + schedulerSQLite RegistryNodes + job stateNode Agent ATelemetry + hardware profileNode Agent BTelemetry + hardware profileDDP Worker APyTorch / NCCLDDP Worker BPyTorch / GlooHeterogeneous consumer nodes coordinated through the cluster-management layer.

AgentForge

Developed a local multi-agent software engineering assistant where specialized AI agents collaborate to analyze requirements, design solutions, critique outputs, and produce implementation plans.

OllamaMulti-Agent SystemsTypeScriptLocal InferenceConsensusQwenGPT-OSS
View Case Study
SYSTEM ARCHITECTURELocal multi-agent software engineering assistantSoftware RequestUser requirementsAnalystRequirements analysisArchitectSystem designCriticReview + validationPlannerImplementation planCode GenerationOllama / local LLMValidated ProjectCode + tests + fixesSpecialized agents collaborate locally with validation, caching, and code-generation workflows.

Optimus: Autonomous EnergyPlus Building Controller

Built a safety-constrained physical-AI controller for autonomous building operations. Uses EnergyPlus simulation with PyEnergyPlus API for live state observation, local LLM (Ollama/Qwen) or deterministic fallback for bounded supervisory planning, and hard safety validation before schedule override injection.

EnergyPlusPyEnergyPlusOllamaQwenStreamlitDockerPhysical AIBuilding Control
View Case Study
SYSTEM ARCHITECTURESafety-constrained EnergyPlus control loopEnergyPlusBuilding simulationSimulation StateTemperature + load + timeAI ControllerLocal LLM / deterministic fallbackSafety ValidatorBounds + constraintsBounded Control ActionSchedule overrideAudit / MetricsTraceable decisionsControl actions remain bounded by explicit safety constraints before being applied to the simulation.

Brain MRI Tumor Segmentation

Investigated multiple 3D medical image segmentation architectures for brain tumor delineation using the BraTS dataset, evaluating baseline and attention-enhanced models.

PyTorch3D U-NetAttention U-NetEfficientNetBraTSMONAIMixed Precision
View Case Study
SYSTEM ARCHITECTURE3D U-Net → EfficientNet → Attention pipelineBraTS 20203D MRI volumesPreprocessingNormalization + patchesBaseline 3D U-NetDice 0.8057EfficientNet U-NetDice 0.8157Attention + EfficientNetDice 0.8256ValidationDice + segmentation metricsBest validation Dice: 0.8256 using the Attention + EfficientNet architecture.

Weather Prediction at Scale

Built a city-specific weather prediction pipeline using gradient boosting models trained on approximately 10 years of historical weather observations across thousands of cities.

XGBoostPythonPandasTime SeriesML Pipelines90GB Data
View Case Study
SYSTEM ARCHITECTURELarge-scale city-specific ML forecastingHistorical Weather90 GB datasetPreprocessingCleaning + feature engineeringCity-Specific ModelsThousands of locationsXGBoostPer-city forecastingFastAPIPrediction servingForecast OutputTemperature / weather predictionsCity-specific XGBoost models are exposed through an API-oriented serving layer.

Speech Emotion Recognition

Built a real-time speech emotion recognition pipeline combining audio preprocessing, MFCC feature extraction, and hybrid CNN-RNN deep learning models for affective computing.

TensorFlowPyTorchLibrosaMFCCCNNRNNReal-time Inference
View Case Study
SYSTEM ARCHITECTUREAudio preprocessing + neural emotion classificationAudio InputSpeech waveformPreprocessingCleaning + augmentationMFCC / FeaturesTime-frequency representationCNNSpatial feature extractionBiLSTMTemporal modellingEmotion PredictionEmotion classesPublished research focuses on data augmentation and lightweight CNN-based speech emotion recognition.

AI Research Orchestrator / Multi-Agent Research Scientist

Developed an autonomous multi-agent research system that plans literature reviews, retrieves academic papers, synthesizes findings, and generates structured research reports.

Node.jsTypeScriptSemantic ScholarOpenAlexLLMsAgent Orchestration
View Case Study
SYSTEM ARCHITECTUREDistributed AI research orchestrationResearch QuestionProblem intakeResearch PlannerQuestion-aware planLiterature RetrievalOpenAlex / Semantic ScholarResearch AgentsAnalyze evidenceExperiment PlannerMetrics + training planDistributed WorkersDMLF / PyTorch DDPResearch BriefEvidence + resultsOffline-demo mode uses deterministic agents and simulated workers where external services are unavailable.