SS
Open to AI Engineering & Research Internships

AI Engineer & ML Systems Builder

I build distributed ML infrastructure, agentic AI systems, and research-driven machine learning applications. Experienced with PyTorch Distributed, LLM orchestration, cloud platforms, and scalable software engineering for real-world AI workloads.

7

Engineering case studies

4,300+

Cities modeled

90 GB

Largest dataset processed

3

Deep learning projects

Selected Engineering Projects

Technical Focus

Distributed ML

  • PyTorch DDP
  • gRPC
  • Distributed Training
  • Scheduling
  • Telemetry
  • NCCL / Gloo

Agentic AI

  • Multi-Agent Systems
  • LLM Orchestration
  • Ollama
  • Tool Use
  • Research Agents
  • Consensus

ML Engineering

  • PyTorch
  • Python
  • FastAPI
  • Docker
  • MLOps
  • Model Serving

Research

  • Computer Vision
  • Medical AI
  • Speech AI
  • ML Systems
  • IEEE Published

Selected Work

View all 7 projects
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.

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

Engineering Highlights

  • Designed node discovery across heterogeneous hardware
  • Managed DDP initialization and rendezvous
  • Implemented a scheduler for available compute nodes
  • Overlap of communication and computation matters more than raw node count
  • A small persistent registry can make a distributed system much easier to debug
  • Telemetry and heartbeat checks are essential in heterogeneous clusters
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.

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

Engineering Highlights

  • Designed role-based agent separation
  • Handled local model execution with Ollama
  • Added caching for repeated reasoning steps
  • Separating agent roles improves clarity of reasoning
  • Local inference is useful when privacy or offline operation matters
  • Consensus steps help catch weak outputs before they reach a user
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.

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

Engineering Highlights

  • Integrated EnergyPlus Python API for live state observation
  • Designed bounded LLM planning with deterministic fallback for offline operation
  • Implemented hard safety validation: PMV/temperature guards (comfort), ITE inlet temperature (data centers)
  • Physical-AI systems require hard safety boundaries, not just ML optimization
  • Deterministic fallback enables evaluation without external LLM dependencies
  • EnergyPlus SQL + JSONL audit trail provides full traceability for building operators
View Case Study

Research & Publications

Speech Emotion Recognition

Real-time speech emotion recognition pipeline with enhanced data augmentation and lightweight CNN. Published at IEEE DECoN 2025.

CNN-RNNMFCCRAVDESSCREMA-DReal-time
Publication

Speech Emotion Recognition: A Human-Centric Framework with Enhanced Data Augmentation and Lightweight CNN

IEEE DECoN 2025

DOI: 10.1109/DECoN67170.2025.11448083

Distributed AI Infrastructure

Scaling training and inference across heterogeneous compute clusters with minimal communication overhead.

PyTorch DDPgRPCSchedulingTelemetry

Medical AI

Applying deep learning to medical imaging for segmentation, classification, and diagnostic assistance.

3D U-NetBraTSAttentionDice 0.8256

Multi-Agent Systems

Designing teams of specialized AI agents that communicate and collaborate on complex tasks.

Agent OrchestrationOllamaConsensusRAG

Large Language Models

Efficient fine-tuning, alignment, and deployment of LLMs for production workloads.

Fine-tuningRAGLocal InferenceAlignment

ML Systems Engineering

Software engineering practices that improve reproducibility, scalability, and deployment of ML applications.

MLOpsDockerFastAPIModel ServingCI/CD

Experience

2025

AI Research Intern

Coding Jr

  • Defined product strategy for 3 core features by synthesizing competitive analysis of 5 major market players and client feedback into actionable requirements
  • Increased client engagement by 15% through 10+ data-driven technical presentations translating complex model outputs into business value
  • Analyzed data from 50+ potential clients to identify user pain points, collaborating cross-functionally with Engineering to prioritize the roadmap
2025

Research Engineer

Independent Research

  • Medical imaging segmentation: BraTS 3D U-Net with Attention mechanisms
  • Speech emotion recognition: CNN-RNN pipeline on RAVDESS/CREMA-D datasets
  • Weather prediction at scale: XGBoost models for 4,500+ cities using 90GB historical data
  • Multi-agent systems: AgentForge and AI Research Orchestrator
  • Published research paper in IEEE
  • Focus on reproducible experimentation, software engineering practices, and practical ML pipelines
2026

Optimus: Autonomous EnergyPlus Building Controller

Honeywell Campus Connect Hackathon

  • Designed safety-constrained physical-AI controller for autonomous building operations using EnergyPlus simulation
  • Integrated PyEnergyPlus API for live state observation and local LLM (Ollama/Qwen) with deterministic fallback for bounded supervisory planning
  • Implemented hard safety validation: PMV/temperature guards for comfort buildings, ITE inlet temperature for data centers
  • Built one-hour bounded cooling relaxation with automatic native schedule restoration
  • Delivered Docker/Streamlit deployment, JSONL audit trail, and EnergyPlus SQL-derived KPI comparison reports
  • GitHub: https://github.com/ShreyanshShakya/Optimus
2026

Participant

OpenAI Build Week

  • Explored Large Language Models, Multi-Agent Systems, AI Infrastructure, Rapid Prototyping, and AI Application Development
  • Built modular AI architectures and collaborative reasoning workflows
  • Developed functional AI prototypes under time constraints
2026

Data Scientist Intern

Springer Capital

  • Built maintainable ML pipelines for business applications
  • Performed exploratory data analysis on structured datasets
  • Designed preprocessing pipelines and developed predictive models
  • Evaluated model performance using appropriate metrics
  • Technologies: Python, SQL, Pandas, NumPy, Scikit-learn, Git

Technical Stack

Distributed ML

PyTorchDDPNCCLGloogRPCCluster ManagementSchedulingTelemetry

LLM Applications

OllamaSemantic ScholarOpenAlexAgent OrchestrationPrompt Engineering

Medical AI

3D U-NetAttention U-NetBraTSDice LossMixed PrecisionMONAInibabelEfficientNet

Speech AI

LibrosaMFCCCNNTensorFlowPyTorchReal-time InferenceRAVDESSCREMA-D

Data Engineering

XGBoostPandasNumPyScikit-learnTime SeriesFeature Engineering90GB Pipeline

Cloud & Infra

AWSDockerSQLiteTelemetrySchedulingEC2BedrockLinuxWSL2GitGitHub

Languages

SQLC++JavaJavaScript

Research & Software Engineering

Literature ReviewExperimental DesignModel BenchmarkingReproducible ResearchTechnical DocumentationSystem DesignModular ArchitectureCode ReviewTestingPerformance Optimization

By The Numbers

90 GB

Largest Dataset Processed

Handled end-to-end in training and evaluation pipelines

4,300+

Cities Modeled

City-specific weather forecasting across a decade of data

7

Engineering Case Studies

Distributed systems, agentic AI, ML, and research projects

1

IEEE Publication

Peer-reviewed research in speech emotion recognition

GitHub Contribution Graph

Let's build something.