SASIKALA N · ML Engineer · 6+ yrs
Mid-LevelAbout SASIKALA
Sasikala N is an AI/ML Engineer with full-stack capabilities, bringing over 6 years of experience in the IT industry. She specializes in deploying AI/ML models into production environments and has strong proficiency in CI/CD for automated builds and deployments. Her expertise extends to orchestrating complex data pipelines and optimizing data processing for large datasets. Sasikala excels in client interactions and requirements gathering, ensuring effective solution development. Her technical prowess is complemented by excellent problem-solving skills and a commitment to teamwork.
Core expertise
Additional skills(24)
Flask
Numpy
Pandas
CircleCI
DockerWhy hire SASIKALA?
Automated build/deployment processes, reducing human intervention by 40%.
Reduced large dataset processing time by 50% using Pyspark.
Created an Airflow orchestrator pipeline, decreasing manual effort by 30%.
Developed an ML model that increased customer base by 20% by improving complaint resolution.
Successfully automated build/deployment processes, reducing human intervention by 40%.
Created an Airflow orchestrator pipeline that decreased manual effort by 30%.
Project highlights(6)
Emotion-Based Recommendation System – Data Scientist
Overview: This project involved building a labeling model to categorize user emotions and a recommendation system based on these emotions. Responsibilities: Paraphrased the training dataset to increase its size. Fine-tuned the BART model to label events based on given emotions. Built a recommendation system using cosine similarity as the metric. Deployed the BART model using AWS Sagemaker for daily predictions.
Key outcomes:
Successfully developed and deployed an emotion-based recommendation system.
Increased training dataset size through paraphrasing.
Vehicle Repair Prediction Model – Data Scientist
Overview: This project focused on developing a machine learning model to reduce unnecessary vehicle repairs caused by sensor failures. Responsibilities: Used Pinecone to find similarity between user input appliances and PDF embeddings. Employed prompt engineering and OpenAI GPT-4 to extract text from matching PDFs. Created Design of Experiments (DOE) for user input using prompt engineering and OpenAI GPT-4. Predicted DOE outcomes and generated DOE graphs using statsmodels and matplotlib.
Key outcomes:
Developed a model capable of reducing unnecessary vehicle repairs.
Utilized advanced NLP and vector search techniques for information extraction.
Customer Complaint Management Model – Data Scientist
Overview: This project created an ML model to identify problematic customer complaints, enabling companies to take quick action to improve customer satisfaction and retention. Responsibilities: Automated build and deployment using CircleCI to minimize human intervention and speed up production processes by 40%. Created an Orchestrator pipeline from data ingestion to data validation using Airflow, reducing manual effort by 30%. Used Pyspark for transforming large datasets, decreasing processing time by 50% compared to Pandas.
Key outcomes:
Automated deployment, speeding up processes by 40%.
Reduced manual orchestration effort by 30% using Airflow.
Decreased data processing time by 50% using Pyspark.
Contributed to a 20% increase in customer base by enabling proactive complaint resolution.
Product Similarity System – Data Scientist
Overview: This project aimed to calculate product similarity using image and text analysis. Responsibilities: Used pretrained models and custom CNN Models to find similarity in images. Generated IMAGE EMBEDDINGS using pre-trained models. Used ANN to select the 5 most similar images. Calculated image similarity using COSINE SIMILARITY Metrics. Used LSH Algorithm to calculate product name similarity.
Key outcomes:
Developed a comprehensive product similarity system using both visual and textual data.
Successfully applied deep learning models for image embeddings and similarity.
FINETUNING BERT MODEL
- This project focused on fine-tuning a BERT model for specific NLP tasks.
- Worked on EDA to reduce missing values.
- Preprocessed input text, converting it to BERT (LLM) model format.
- Created a metric function using PyTorch to calculate accuracy and F1 score.
Key outcomes:
Successfully fine-tuned a BERT model for specific tasks.
Implemented custom metric functions for performance evaluation.
6+ years of industry experience
- •Emotion-Based Recommendation System— Data ScientistPython · Flask · Numpy · Pandas +6
- •Vehicle Repair Prediction Model— Data ScientistPython · Azure · LangChain · Statsmodels +4
- •Customer Complaint Management Model— Data ScientistPyspark · CircleCI · Docker · Airflow +3
- •Product Similarity System— Data ScientistPython · Flask · Microservice Architecture · Numpy +4
- •FINETUNING BERT MODELNLP · PyTorch · Azure · React.js
- •PRODUCT SIMILARITYPython · Flask · Microservice Architecture · Numpy +4
- •Emotion-Based Recommendation System— Data ScientistPython · Flask · Numpy · Pandas +6
- •Vehicle Repair Prediction Model— Data ScientistPython · Azure · LangChain · Statsmodels +4
- •Customer Complaint Management Model— Data ScientistPyspark · CircleCI · Docker · Airflow +3
At a Glance
Single contract. Billed in USD.
Typically responds within 4 business hours.
Top Skills
Vetted by WitaristNot sure if this is the right fit?
Tell us your requirements and we'll match you with the best candidates.
SASIKALA N
Python Developer