Avnish Singh
Open to work1 years experienceMachine Learning Engineer
Machine Learning Engineer and applied ML researcher.
Standing Rank
Rank Not Available
Developer Badges
No badges earned yet
Skills & Technologies
25 skillsWork Experience
Junior Research Fellow - Machine Learning Researcher
National Institute of Technology Raipur
- Conducted applied machine learning research on Advanced Persistent Threat detection for Industrial IoT systems using graph-based representation learning, federated learning, and host-level security data.
- Developed a graph-based APT detection framework using provenance-graph construction, graph attention-based node embeddings, and a stacking ensemble, achieving 99.90% F1-score and 0.01% false-positive rate on the DARPA E3 Theia dataset.
- Built data-processing and experimentation pipelines for provenance-graph construction, representation learning, model training, benchmarking, and performance evaluation using Python, PyTorch, and scikit-learn.
- Developed a federated threat-detection pipeline using Flower, distributed model training, and server-side model aggregation; evaluated configurations with 3, 5, 7, and 10 clients over 10 communication rounds on the Edge-IIoT dataset.
- Achieved up to 98.79% accuracy and 98.14% F1-score across evaluated federated configurations while keeping client data decentralized.
Projects
Document Portal
- Developed an LLM-powered document intelligence application supporting multi-document question answering, single-document analysis, and document comparison.
- Implemented document-processing and LLM orchestration workflows using LangChain with Google, Groq, and Hugging Face models.
- Containerized and deployed the application on Google Kubernetes Engine with health checks, autoscaling, resource management, and Kubernetes secrets.
- Automated cloud infrastructure provisioning and application deployment using Terraform and GitHub Actions.
Data Insight AutoGen
- Developed an agentic data-analysis application that enables users to upload CSV datasets and request analysis through a conversational interface.
- Orchestrated AutoGen agents to generate and execute Python workflows for automated data exploration, analysis, and visualization.
- Executed AI-generated code inside isolated Docker containers to separate execution from the application environment and improve reproducibility.
End-to-End MLOps Pipeline
- Built an end-to-end MLOps workflow covering cloud data ingestion, preprocessing, feature storage, model training, inference, drift detection, and monitoring.
- Orchestrated data ingestion from Google Cloud Storage using Apache Airflow and persisted processed data in PostgreSQL.
- Integrated Redis for feature storage and Alibi Detect for monitoring changes in incoming feature distributions.
- Exposed model predictions through a Flask application and collected prediction and drift metrics using Prometheus and Grafana.
Graph-Based EEG Classification
- Developed an EEG processing pipeline covering signal preprocessing, segmentation into 5-second and 8-second epochs, time- and frequency-domain feature extraction, and graph construction from functional relationships between EEG channels.
- Implemented and trained Graph Convolutional Network (GCN) and hybrid GCN-LSTM models to capture spatial relationships between EEG channels and temporal signal patterns for schizophrenia classification.
- Evaluated model performance using multiple random seeds and 5-fold cross-validation, achieving 99.25% accuracy, 99.24% F1-score, and 99.20% AUC with the GCN-LSTM model on 8-second EEG epochs.
Leaderboard Standings
Leaderboard Position Pending
Global test scores, peer standing percentiles, and algorithm leaderboard ranks are updated dynamically.
Assessment Highlights
Assessments Not Completed
Coding evaluations, system assessment results, and conceptual score badges will appear here after taking a test.
AI Collaboration Score
AI Collaboration Score Pending
Developer coding behavior, assistant cooperation, and AI pair-programming indicators are evaluated during live coding sessions.
Role Compatibility Profile
Role Compatibility Analysis Pending
Custom matching reports, candidate role compatibility percentiles, and core engineer strength profiles are processed once conceptual code screenings are complete.
Achievements
Research Publication
Gosala, Bethany, Avnish Ramvinay Singh, Himanshu Tiwari, and Manjari Gupta. “GCN-LSTM: A hybrid graph convolutional network model for schizophrenia classification.” Biomedical Signal Processing and Control 105 (2025): 107657.
Research Publication
Singh, Avnish, and Govind P. Gupta. “Foundations of edge intelligence.” In Edge Intelligence and Analytics for Internet of Things, pp. 35–43. CRC Press, 2025.
Research Publication
Singh, Avnish Ramvinay, and Govind P. Gupta. “Advanced Persistent Threats Detection Framework for Industrial-IoT System Using Graph-Based Autoencoder and Stacking-Based Ensemble Technique.” In International Symposium on Artificial Intelligence, pp. 459–468. Springer Nature Singapore, 2025.
Research Publication
Singh, Avnish Ramvinay, and Govind P. Gupta. “Fed-NODE: Federated Learning based Threat Detection Framework for Edge-based Industrial Internet of Things.” MIND 2025, MNIT Jaipur.
About Details
Professional Bio
Machine Learning Engineer and applied ML researcher with experience developing graph-based threat detection systems, federated learning frameworks, GenAI applications, and end-to-end MLOps pipelines. Published researcher with hands-on experience across data processing and model development.
M.Sc. in Computational Science and Applications (Data Science) in Data Science
Banaras Hindu University (2022 - 2024)
B.Sc. in Computer Science in Computer Science
University of Mumbai (2019 - 2022)