卍 B A N K A I 卍
SUBSTITUTE SHINIGAMI
霊圧検知 · Available for Opportunities

SAYANTAN SIKDAR

— DATA SHINIGAMI —

K654321`

AI/ML ENGINEER · DATA SCIENTIST · IEEE RESEARCHER

I

Kolkata, India

0+Years Experience
0+Projects Completed
0IEEE Publication
0+Certifications

ABOUT ME

CS postgraduate student specializing in AI and ML, with hands-on industry experience in building scalable ML solutions, time-series forecasting, LLM orchestration, and production optimization. Published researcher with IEEE recognition. Passionate about turning complex data into actionable insights and building intelligent systems that make a real-world impact.

TECHNICAL SKILLS

Technologies and tools in my Zanpakutō arsenal

Languages

PythonSQLJava

ML/DL Frameworks

PyTorchTensorFlowKerasScikit-learn

Time-Series

NeuralForecastN-BEATSN-HiTSTFTDartsProphetXGBoost

LLM & Agentic AI

LangChainCrewAILlamaIndexDSPyAutoGenPrompt EngineeringContext EngineeringLoRA/QLoRARAG

Optimization

CP-SATGoogle OR-ToolsBayesian OptimizationCVFS-CMA-ESTuRBOSAASBO

Data Tools

PandasNumPyMatplotlibOpenCVNLTKAirflowDatabricksDVC

Cloud & DevOps

AWS SageMakerS3GlueGCPDockerGitCI/CDPrometheusGrafana

Web

FastAPIFlaskReactREST APIs

MISSION BRIEFINGS

Soul Society classified project archives

Advanced Basket Trading Optimization Framework

Implemented and benchmarked 6 optimization algorithms (CVFS-CMA-ES, Bayesian Optimization, TuRBO, SAASBO) for cointegration-based basket trading. Achieved 3.851 Sharpe ratio with CVFS-CMA-ES and 63.76% total return with Bayesian Optimization.

PythonPyTorchBayesian Opt.Time-SeriesCP-SAT

Book Recommendation System with MLOps

End-to-end MLOps book recommender using collaborative filtering with automated data pipelines, Airflow orchestration, DVC versioning, FastAPI services, React frontend, Docker deployment, and Prometheus/Grafana monitoring. Reduced manual operations by 80%.

PythonAirflowFastAPIReactDockerDVC
卍 IEEE PUBLISHED

Fruit Grading using Deep Learning

Designed DNN framework with AlexNet and ResNet-v2 for automated fruit grading, achieving 99% accuracy on 2.2K images and 98.33% accuracy with 0.9752 F1-score on 15.5K images.

TensorFlowKerasOpenCVDeep LearningCV

YOLOv9n Automatic Meter Reading

Deep Learning-based AEMR system using YOLOv5–v8 achieving 99% accuracy, 94% precision, 93% mAP. Applied erosion and HSV correction on 466 images, reducing errors by 95% and saving 500+ hours annually.

PythonYOLORoboflowOpenCV

Morsaabs Restaurant Website

Engineered full-stack restaurant website using 8 strategic AI prompts — compressed 90+ hours of development into 4.5 hours, delivering at 70% lower cost than traditional developers.

FastAPIReactFlaskClaude CodeCursor AI

BATTLE HISTORY

Professional journey through the divisions

AI/ML Research Intern

CDAC Kolkata
Oct 2024 – Mar 2025
  • Fine-tuned OpenNMT and integrated LLMs for Neural Machine Translation, improving BLEU scores by 18% and reducing errors by 22%
  • Engineered synthetic data pipelines processing 50K+ sentence pairs with Airflow orchestration
  • Enhanced model performance through hyperparameter optimization

Machine Learning Trainee

WBSEDCL
Mar 2024 – Apr 2024
  • Developed YOLOv5–v8 based AEMR system achieving 99% accuracy, 94% precision, 93% mAP
  • Applied erosion and HSV correction on 466-image set, reducing errors by 95% and saving 500+ hours annually

AI Research Intern

NIELIT Kolkata
Sep 2023 – Mar 2024
  • Designed DNN framework with AlexNet and ResNet-v2 for automated fruit grading
  • Achieved 99% accuracy on 2.2K images and 98.33% accuracy with 0.9752 F1-score on 15.5K images

TRAINING GROUNDS

M.Tech (IT)

Netaji Subhas University of Technology

Jul 2025 – Jun 2027 · CGPA: 6.80

Advanced DSA, Machine Learning, Deep Learning, NLP, Blockchain

B.Tech (CSE - AI&ML)

Haldia Institute of Technology

Oct 2020 – Jul 2024 · CGPA: 8.65

Time-Series Analysis, Probability & Statistics, Software Engineering, DSA, DBMS, OS

Google Cloud Training: ML-AI, DevOps & Kubernetes, Cloud Engineering
AWS Machine Learning Scholarship 2021, Udacity
GATE CS 2024 — Qualified
Prompt Engineering & LLM Fine-tuning (Self-directed)

RESEARCH SCROLLS

卍 IEEE

Fruit Grading using Deep Learning

Sayantan Sikdar et al.

Designed DNN framework with AlexNet and ResNet-v2 for automated fruit grading, achieving 99% accuracy on 2.2K images and 98.33% accuracy with 0.9752 F1-score on 15.5K images.

IEEE International Conference 2024
Open Senkaimon · IEEE Xplore

SOUL CONTACT

Send a Hell Butterfly my way