Hi, I'm Akanksha👋

I'm an AI Engineer, and Data Scientist with a passion for blending state-of-the-art technology with genuine human impact—always with a smart seasoning of business rules and real-world context. Armed with a Master's in Artificial Intelligence and Business Analytics from the University of South Florida, I've built my career around solving real-world problems—starting from analyzing challenges and gathering requirements to developing ML, DL, and Generative AI models, engineering cloud data pipelines, and deploying next-generation AI solutions in the cloud.

I love crafting intelligent products with tools like LangChain, Hugging Face Transformers, Vision Transformers, advanced LLMs, dynamic RAG pipelines, and seamless API integration—bringing it all to life on cloud platforms like AWS SageMaker, rolling out real-time endpoints and scalable deployment with Docker containers. 🚀☁️🐳

Whether it's orchestrating multi-agent systems, fine-tuning LLMs with QLoRA, or implementing MLOps best practices, I thrive on building robust, explainable AI that's truly ready for business. I design MCP Autonomous Agents with OpenAI, build agentic RAG pipelines with FAISS, ChromaDB, and AstraDB, and engineer multi-agent frameworks with Strands,CrewAI, LangGraph, and AutoGen—where specialized agents collaborate, reason, and delegate tasks.

In Machine Learning, I've delivered production models for churn prediction, fraud detection, pricing optimization, and sentiment analysis using Random Forest, XGBoost, and Gradient Boosting. On the Deep Learning side, I've built advanced classification models with CNNs, RNNs, and Vision Transformers.

When I'm not exploring new tech, you'll find me behind a camera 📷 or sharing creative stories online— because I believe every great solution starts with a spark of curiosity and a little creativity. ✨

Akanksha Kushwaha headshot

Skills

I specialize in transforming raw data into actionable insights using state-of-the-art AI, multi-agent orchestration, LLM fine-tuning, and modern MLOps. My toolkit spans everything from classical ML and deep learning to LLMs, cloud deployments, and advanced vector databases.
sql-logo SQL
python-logo Python
pandas-logo Pandas
numpy-logo Numpy
powerbi-logo Power BI
tableau-logo Tableau
matplotlib-logo Matplotlib
seaborn-logo Seaborn
scikit-logo Sci-kit Learn
tensorflow-logo Tensor Flow
azure-logo Azure
aws-logo AWS
langchain-logo LangChain
huggingface-logo Hugging Face
docker-logo Docker
openai-logo OpenAI
strands-logo Strands
ollama-logo Ollama
streamlit-logo Streamlit
gradio-logo Gradio
anthropic-logo Anthropic
faiss-logo FAISS
chroma-logo Chroma
awsvs-logo Amz S3 Vectors

Experience

Here's a summary of my professional experience.

Gen AI Engineer Volunteer

American Red Cross, NHQ:CIO, USA

June 2024 – Present

  • Autonomous Incident Response: Built multi-agent cloud platform for the American Red Cross with AWS Bedrock, LangChain, LangGraph & JIRA, cutting MTTR by 45%, manual handling by 40%, and alert fatigue by 42%.
  • AI-Driven Product Pricing: Engineered pricing system with Hugging Face Transformers, QLoRA (LLaMA-8B) & ML baselines, reducing errors by 62% and boosting accuracy to 70%+.
  • AskMyPDF RAG Assistant: Developed RAG tool for the American Red Cross using Bedrock (Claude & Llama 3), Titan, FAISS & Streamlit, enabling natural language queries on policy PDFs for faster insights.

Senior Data Engineer

Archents India Pvt ltd, India.

July 2019 – March 2022

  • Azure Data Migration: Migrated on-prem data warehouse to Azure PaaS, automating pipelines for faster data refreshes (14-hour improvement) and scalability.
  • Predictive ML Models: Built Python/scikit-learn/TensorFlow models for demand forecasting and risk detection, improving accuracy by 65% and reducing delays.
  • ETL Optimization: Engineered scalable ETL pipelines (SQL Server, Oracle, Azure) using PolyBase, improving data quality and traceability.
  • BI Dashboards: Created Power BI/Tableau dashboards with DAX and KPIs, speeding up decisions and reducing report latency by 12 hours.

Data Engineer

AL-Ameen Project Implementation Authority, India

April 2017 – July 2019

  • SQL Database Optimization: Engineered and managed high-performance SQL databases, improving query speeds and reliability.
  • Client Onboarding: Streamlined new client integration by setting up AD security groups and ownership roles.
  • Security & Access Control: Implemented database roles and AD group policies, enhancing security and accessibility.
  • Automation: Developed SQL Stored Procedures and Functions to automate tasks, increasing team efficiency.
  • Database Design: Built schemas, tables, indexes, and views to support complex data needs.

Assistant Engineer Purchase

ACON PEB INDUSTRY Pvt. Ltd, India

January 2015 - April 2017

  • Procurement Management: Streamlined mechanical commodities procurement via Salesforce ERP and Excel, cutting requisition-to-order time to 2 weeks through logistics coordination.
  • Invoice Automation: Developed VBA scripts to automate vendor invoice reconciliation, speeding up payment clearance with finance/warehouse teams.
  • Cost Optimization: Negotiated payment terms and leveraged ERP insights to reduce costs while maintaining quality.
  • ERP Record-Keeping: Maintained accurate records of POs, invoices, and goods receipts in the ERP system.

Projects

Here are some of the projects I've worked on. You can explore them on my GitHub profile.
Project 0 AI-Powered Financial Banking Assistant

AI-Powered Financial Banking Assistant

- Developed a secure LLM-based assistant to mitigate prompt injection, jailbreaking, and data leakage in financial apps.

- Used AWS Bedrock Guardrails for PII redaction and dynamic role-based input/output filtering.

- Implemented a multi-agent system with semantic routing via Strands Agent Framework for better query handling.

- Built hybrid RAG using FAISS, Titan embeddings, PyPDF, and Notion FAQs for semantic answer accuracy.

View on GitHub
Project 1 Autonomous Multi-Agent JIRA Incident Response System

Autonomous Multi-Agent JIRA Incident Response System

- Built with Amazon Bedrock, LangChain, LangGraph, OpenAI embeddings, Retrieval-Augmented Generation (RAG), Python.

- Integrated AWS CloudWatch, CloudTrail, JIRA API, LangSmith.

- Reduced mean time to resolution, manual ticket handling, and alert fatigue.

- Deployed to AWS SageMaker for scalable model inference.

View on GitHub
Project 2 Multi-Agent Trading Simulator using MCP

Multi-Agent Trading Simulator using MCP

- Used OpenAI agentic orchestration, Anthropic MCP microservices.

- Modular MCP servers (Market, Account, Push).

- Python, Gradio UI, Polygon.io, Pydantic, Asyncio, SQLite, Docker for deployment, guardrails for compliance.

- Custom logging/tracing with LogTracer, Pushover notifications.

- Showcased agentic automation for banking/payment operations.

View on GitHub
Project 3 AI-Amazon Product Pricing

AI-Amazon Product Pricing Model

- Built an end-to-end pricing pipeline using Python, Hugging Face Transformers, and PEFT.

- Cleaned and prepped millions of listings with pandas and prompt engineering.

- Benchmarked ML, DL, and LLMs, fine-tuning LLaMA 8B with QLoRA and bitsandbytes.

- Cut prediction error by 62% and boosted accuracy with model quantization. Debugged with Weights & Biases and deployed APIs for real-time results.

View on GitHub
Project 4 Autonomous Analytics Engine for Banking

Autonomous Analytics Engine for Banking

- Modular agent-as-tools architecture (Strands SDK + AWS Bedrock: Claude & Nova).

- Orchestrated Lead, Fundamental, Technical, and Market Analyst agents.

- Real-time financial data APIs, technical indicator calculators, dynamic code execution, secure shell access.

- Human-in-the-loop controls and compliance gates (guardrails).Langfuse for observability/debugging.

View on GitHub
Project 5 Machine Learning: Predicting Parkinson's Disease

Machine Learning: Predicting Parkinson's Disease

- Developed supervised ML models: Logistic Regression, SVC, Decision Tree, Random Forest, Gradient Boosting, Ensemble.

- Performed extensive feature engineering and applied PCA for dimensionality reduction.

- Used cross-validation and hyperparameter tuning for robust model selection.

- Achieved a 94% improvement in macro F1 score with ensemble learning.

- Delivered reliable multi-class classification for disease progression.

View on GitHub
Project 6 Deep Learning: Human Emotion Recognition

Deep Learning: Human Emotion Recognition

- Benchmarked LeNet, ResNet-34, EfficientNetB4, VGG-16, Vision Transformer, Hugging Face ViT for facial emotion classification.

- Automated data wrangling, advanced augmentation (Keras, Albumentations), and class balancing.

- Improved interpretability using GradCAM, feature map visualization, and transformer patch attention.

- Managed experiments and MLOps with Weights & Biases.

- Deployed best model for real-time inference via FastAPI for scalability.

View on GitHub
Project 7 Power BI: Impact of Unemployment on Mental Health

Power BI: Impact of Unemployment on Mental Health

- Designed an interactive Power BI dashboard visualizing links between unemployment rates and mental health outcomes (depression, suicide rates).

- Integrated and transformed data from World Bank, CDC, WHO, and OECD for regional, temporal, and demographic analysis.

- Built dynamic maps, trend lines, and drill-through reports to uncover disparities and trends across regions and populations

- Developed custom DAX measures to compare mental health impacts across socioeconomic groups.

View on GitHub

TESTIMONIALS

Feedback from clients and collaborators I've had the privilege to work with.

Justin Larson

"Beyond her technical excellence, Akanksha is incredibly responsive and a genuine pleasure to collaborate with. Her communication is clear and proactive, and she brings a positive, solutions-oriented mindset to every interaction. Whether navigating complex requirements or adapting to shifting priorities, she remains composed, thoughtful, and focused on delivering value. Akanksha’s combination of professionalism, technical skill, and collaborative spirit makes her stand out. I highly recommend her."

Justin Larson

Owner, Hamelin LLC/ XPO Principal Solutions Architect, USA
Koyal Roy

"Akanksha is very driven and detail oriented. She was instrumental in developing multiple capabilities in various data pipelines, its optimization exercises . She has a keen sense of understanding the underlying business , processes that gets reflected in her discussions and work deliverables. She was one of the critical resources in the team handling complex architectural developments. I really enjoyed working with Akanksha and the collaboration we shared. I wish only best for both her professional & personal endeavors."

Koyal Roy

Cloud Data Engineer, Mondelēz International
Siva Barinepalli

"I had the pleasure of working with Akanksha, and I can confidently say she is a dedicated and talented professional. Her strong problem-solving skills, proactive attitude, and collaborative nature made a significant impact on our team's success. She consistently delivered high-quality work and was always willing to go the extra mile to support her teammates. Akanksha would be a valuable asset to any team, and I highly recommend her for any opportunity she pursues.I wish best for all her future endeavors."

Siva Barinepalli

Azure Data Engineer, Archents India Pvt Ltd
Sai Krishna Sanem

"I had the pleasure of working with Akanksha, and she is truly an outstanding professional. She goes the extra mile to achieve results, ensuring every detail is taken care of. Her precise understanding of complex problems and exceptional problem-solving skills truly set her apart. She is good a team player and I highly recommend her to anyone looking for a reliable and talented professional."

Sai Krishna Sanem

Senior Data Engineer, LHS Property Services Ltd, UK
Raj Kaur

"Akansha is an excellent professional to work with, very knowledgable and an all-rounder. She is a great asset to the company in Azure Cloud, AI and ML related emerging technologies. I wish her all the very best professionally and personally to make good decisions and soar in her life ! "

Raj Kaur

Senior Solutions Engineer, XPO USA