Charitha p

AI/ML Engineer @ Second Nature

About

Currently contributing as a Senior Software Engineer at Second Nature Brands, specializing in building scalable ML inference runtimes. Core competencies include optimizing Transformer models (CNNs, LSTMs) and enhancing inference efficiency through quantization, pruning, and Kubernetes-native deployment architectures. At Second Nature Brands, developed multimodal model-serving pipelines using TorchServe and vLLM, reducing latency and improving object detection accuracy. Collaborated on real-time crime prediction systems, leveraging MobileNetV2 optimizations. Passionate about advancing ML infrastructure to deliver high-availability solutions while fostering innovation and teamwork within the organization.

Country

United States

City

Virginia Beach

Industry

Information Technology & Services

Skill

Anomaly Resolution, NLP Libraries, BERT (Language Model), GPT-4, OpenVINO, quantization, Distillation, TGI, MULTI - CLOUD ML, AZURE ML, KUBERNETS, Grafana, PROMETHUES, Elastic Stack (ELK), PyTorch, CNN Pathfire, lstm, Transformers, Torch Serve, VLLM

Experience

AI/ML Engineer

Second Nature

2024-7 - 2025-4 · 10 mos

North Carolina, United States

• Collected, cleaned, and preprocessed structured and unstructured datasets, improving data quality for ML models. • Conducted EDA to identify trends, patterns, and anomalies, driving feature engineering and model selection. • Built, validated, and deployed ML models (regression, classification, clustering, NLP, time series) in production. • Developed end-to-end ML pipelines using Python, TensorFlow, PyTorch, and Scikit-learn, reducing deployment time by 30%. • Implemented model monitoring, logging, and versioning using MLflow for reproducibility and performance tracking. • Designed dashboards with Tableau and Power BI to communicate actionable insights to leadership and product teams. • Experimented with generative AI and LLM models for internal automation tools, exploring AI-first workflows. • Collaborated with data engineers to maintain scalable, reliable data pipelines for batch and real-time processing.

Data Scientist

Cummins Inc.

2023-8 - 2024-6 · 11 mos

Brooklyn, New York, United States

• Designed and implemented a hybrid ML pipeline integrating DistilBERT embeddings with gradient-boosted decision trees, improving accuracy by 24% and reducing computational overhead by 30%. • Implemented NLP pipelines using CoreNLP, SpaCy, and Whisper for multi-modal conversational analytics. • Researched SLU methodologies, focusing on transformer architectures for intent recognition and slot filling. • Leveraged causal tracing, NNsight, and knowledge graphs for model interpretability in large language models. • Contributed to ONR project on team optimization using RL and multi-agent systems, improving performance by 22%.

Machine Learning Engineer

Infosys

2021-7 - 2023-12 · 2 yrs 6 mos

Hyderabad, Telangana, India

• Leveraged frequent pattern mining to segment e-commerce customers based on browsing and purchase behavior. • Improved client customer retention by 18% through personalized product recommendations. • Increased average order value by 12% among segmented customer groups using advanced ML techniques. • Collaborated with cross-functional teams to integrate ML models into production systems, ensuring scalability and performance. • Developed Hola Enterprise, a SaaS platform for customizable video generation, increasing user engagement by 30%. • Engineered modular web components, reducing video rendering time by 25% and boosting adoption rates. • Implemented advanced designer features, improving UX scores by 45% and decreasing design time by 35%. • This project was awarded with Top 50 Emerging Startups Projects from India by NASSCOM.