|
|
|
class ShashankMankala:
def __init__(self):
self.role = "AI Engineer & Data Scientist"
self.location = "New York, NY | Open to Relocation"
self.education = "M.S. Data Science @ SUNY Buffalo"
self.open_to = ["Full-Time Roles", "Relocation", "Interesting Problems"]
self.working_on = "Optimizing AI tokens usage"
@property
def expertise(self):
return {
"Multimodal AI" : ["FashionCLIP", "Gemini Embeddings", "KNN Retrieval", "Vision-Language Models"],
"Forecasting" : ["Temporal Fusion Transformer", "Multi-Horizon TFT", "Probabilistic Forecasting"],
"LLM Systems" : ["RAG Pipelines", "Custom Decoding", "Inference Optimization", "Prompt Engineering"],
"Agentic AI" : ["LangGraph", "CrewAI", "Tool-Use", "Multi-Agent Orchestration"],
"Generative AI" : ["Diffusion Models", "LoRA / QLoRA Fine-tuning", "RLHF", "Synthetic Data Generation"],
"MLOps" : ["Docker", "Kubernetes", "MLflow", "Airflow", "CI/CD", "Feature Stores"],
"Data Engineering" : ["Snowflake", "Spark", "Databricks", "ETL Pipelines", "Streaming ML"],
"Vector & Retrieval" : ["FAISS", "Pinecone", "Weaviate", "Embedding Search", "Reranking"],
}
def current_mission(self) -> str:
return "Bridging cutting-edge ML research with production-ready systems."|
Data Scientist Intern · Feb 2026 - Present · New York, NY
|
Data Scientist Intern · Sep - Dec 2025 · Buffalo, NY
|
|
Lead Data Analyst · Sep 2023 - Jul 2024 · Bangalore, India
|
Software Engineer Intern · Jun 2022 – Jun 2023 · Hyderabad, India
|
M.S. Data Science
State University of New York at Buffalo
Aug 2024 - Dec 2025
B Tech. Computer Science & Engineering
Lovely Professional University
Aug 2019 - May 2023
|
Stack: A distributed, async video processing pipeline for long-form YouTube content — background ingestion, job tracking, and fault recovery. 🔑 Key challenges solved:
|
Stack: A local LLM inference engine with full logits-to-decoding pipeline supporting grammar-constrained generation and streaming. 🔑 Key features:
|
- Builds notebooks + Builds production pipelines
- Stops at model accuracy + Owns deployment, monitoring & retraining
- Works with one modality + Multimodal: vision, language, time-series
- Uses pre-built APIs + Implements custom decoding from scratch
- Data Scientist OR Engineer + Both end-to-end, from raw data to inference
| 40,000+ | 2 | 3 | 4 | 30% |
|---|---|---|---|---|
| SKUs forecasted in production | Custom ML systems built from scratch | Cloud platforms shipped on | Automation frameworks built | Reduction in manual QA effort |
