Machine Learning Engineer

Srilakshmi Kanagala

Boston,MA 857-***-**** ********.*@************.*** linkedin.com/in/srilakshmikanagala/ github.com/kanagalasrilakshmi Work Experience

Machine Learning Engineer, Dassault Systems, Providence, RI May 2023 - Dec 2023

• Spearheaded development of RAG chatbot, reducing regulatory compliance penalties by $68K annually and strategically increased user adoption from 10% to 78% in 7-weeks.

• Optimized search efficiency by 40% by storing Mistral-7b embeddings in OpenSearch vector db and utilizing Llama-2 for response generation, achieving 0.68 ROUGE evaluation metric score.

• Built an ETL training pipeline to perform data mining of 30 GB unstructured healthcare data, stored in Snowflake using Spark.

• Pioneered GNN-based supervised machine learning model for stress location classification, achieving 96% accuracy.

• Built a CI/CD pipeline in AWS for scalable model development and storage, expediting product design analysis by 50%. Machine Learning Engineer, Qualcomm, Hyderabad, India Jan 2021 - Aug 2022

• Deployed BERT model for feedback analysis, uncovering roadblocks and resolved problems, cutting service requests by 20%.

• Reduced request completion time by 167 hours through automation of data preprocessing and back testing using Jenkins CI/CD pipeline.

• Mitigated $50K potential losses by enhancing anomaly detection efficiency using SVM and XGBoost, A/B testing and statistical analysis.

• Collaborated with business stakeholders to deploy ML tracking tool, reducing response time by 50% by automating processing, anomaly detection and status tracking.

• Created agile KPI dashboards for advanced data analysis using AWS Quick Sight to visualize data anomalies; reducing detection and decision-making time by 12 hours.

Data Scientist, Vitra.ai, Bengaluru, India Oct 2020 – Dec 2020

• Engineered NLP pipeline to summarize lengthy lectures with 85.63% accuracy, cutting down time spent by 50%[paper link].

• Led a team of 4 in building article scraping tool using Selenium, BS4 and SQL that reduced article search time by 45%.

• Boosted performance and client engagement by 35% by creating interactive product interface using ReactJS with Django.

• Amplified annual customer retention by $30K and elevated user satisfaction by 70% through sentiment analysis and Tableau visualization. Data Scientist, Fulcrum GT, Hyderabad, India Dec 2019 - Sep 2020

• Predicted client churn using Random Forest & Logistic Regression with 83% accuracy; enhanced customer retention understanding 10x.

• Launched interactive Power BI dashboard for visualizing financial trends and client retention metrics; reducing analysis time by 20 hours.

• Deployed dashboard using Docker, Amazon ECR, AWS EKS and Kubernetes that elevated performance, reliability, accessibility by 60%.

Skills

Programming Languages : Python, SQL, C, C++, Java, JavaScript, Matlab, HTML, CSS, R. Data Science Tools : PyTorch, Keras, TensorFlow, Scikit-Learn, Numpy, Pandas, OpenCV, Matplotlib. NLP Tools : Spacy, NLTK, TextBlob.

Gen AI tools : HuggingFace, Langchain.

Databases: PostgreSQL, MySQL, NoSQL, Oracle, SQLite. Cloud: AWS - QuickSight, EC2, Sage Maker, DynamoDB, Lambda, OpenSearch, Azure Data bricks, Snowflake, Spark. Software Tools: Github, Gitlab, Jenkins, Tableau, PowerBI, Pinecone, Milvus, Faiss, Docker, Kubernates, NodeJS, Jira. Projects

Prediction of Customer Lifetime Value (CLV) XGBoost, Random Forests, Feature importance analysis

• Improved CLV prediction accuracy by 14.7% over baseline by training and fine-tuning of diverse regression models: Linear Regression, Decision Tree, Random Forest, and XGBoost.

• Extracted key predictors of CLV through feature importance analysis, to drive marketing strategies and bolster customer retention. Creation and Mining of a Medical Database MySQL, SQL, AWS RDS, Databases, Normalization

• Designed a logical data model, normalized relational schema and loaded 350,000+ records into MySQL database on AWS utilizing R.

• Enhanced query retrieval time by 26% leveraging star and snowflake schemas to build summary and fact tables. Time series forecasting with DeepAR and Temporal Fusion Transformer LSTM, DeepAR, Prophet, TFT.

• Forecasted energy consumption leveraging DeepAR, Prophet and TFT models, adjusting parameters to maximize forecasting accuracy.

• Outperformed traditional ARIMA and LSTM models by over 12.7% by implementing multi-step ahead forecasting, delivering actionable insights for efficient resource management.

Education

Northeastern University, Boston MA Sep 2022 - Dec 2024 Master of Science in Artificial Intelligence GPA: 4.0/4.0 Coursework: Machine Learning, HCI, Large Language Modeling, Algorithms, Pattern Recognition, MLOPs, Computer Vision Vellore Institute of Technology, Chennai, India Jul 2016 - Jun 2020 Bachelors in Electronics and Computer Engineering GPA: 9.04/10.0 Coursework: NLP, Database Management Systems, Data Analytics and Visualization, Computer Vision, Statistics

Contact this candidate