Natural Language Processing (NLP)
Designing text pipelines, retrieval-augmented generation (RAG), classification models, and guardrails for reliable, production-ready applications.
Build & ship ML that matters —
Aspiring Data Scientist & Machine Learning Engineer passionate about turning data into measurable impact. Experienced in developing NLP pipelines, forecasting models, recommender systems, and deploying production-grade ML solutions with MLOps and AWS. Strong foundation in data engineering and visualization for clear, actionable insights.
NLPForecasting
RecSysMLOpsGenAI
Ramesh Shrestha
Data Scientist & ML Engineer
I am a Software Engineering (AI) graduate and aspiring Data Scientist & Machine Learning Engineer, currently seeking opportunities to apply my skills in real-world projects. I am passionate about building end-to-end ML solutions — from data preparation and feature engineering to model training, deployment, and monitoring — with a focus on scalability, reliability, and measurable outcomes.
My expertise spans Python, SQL, C++, and PySpark for data science; TensorFlow, PyTorch, and XGBoost for modeling; and AWS, Docker, and CI/CD for cloud & MLOps. I also bring strengths in NLP, forecasting, recommender systems, and deep learning, alongside experience in visualization tools such as Tableau, Power BI, and Streamlit to deliver actionable insights.
My objective is to contribute to teams where I can solve business problems with data-driven solutions, improve decision-making through ML systems, and continue growing into a well-rounded Data Scientist & ML Engineer.
Location
Sydney, NSW, Australia
Work Rights
Full-time in Australia
Focus Areas
Time-Series Forecasting
Building probabilistic and deep learning models (e.g., TFT) with disciplined feature engineering to improve demand planning and reduce error rates.
Recommender Systems
Developing collaborative and session-based models, leveraging feature stores and evaluation frameworks to boost engagement and CTR.
MLOps & Cloud
Deploying ML models with Docker, FastAPI, and CI/CD; monitoring performance in production and scaling with AWS (SageMaker, Glue, Redshift).
Data Engineering
Building reliable pipelines with Pandas, PySpark, and dbt; ensuring clean, high-quality data for analytics and ML systems.
Visualization & Storytelling
Turning data into decisions with Tableau, Power BI, and Streamlit; delivering insights through clear, impactful narratives.
Skills
End-to-end ML with a product mindset — depth in modeling, breadth across data & MLOps.
Machine Learning
NLP (RAG, cls.)Time-SeriesRecommenders
PyTorchTensorFlowscikit-learn
XGBoostFaissCross-encoders
Cloud & MLOps
AWS (SageMaker, S3, Glue, Redshift)
DockerFastAPI
CI/CD (GitHub Actions)MonitoringGuardrails
Data Engineering
PandasNumPyPySpark
dbtSQL (RDBMS)Feature Store
Languages
- Python
- SQL
- C++
Analytics & Visualization
TableauPower BIStreamlit
A/B readoutsStorytelling
Experimentation & Evaluation
Offline/Online evalA/B testing
MAPE / F1 / ROC-AUCLatency & cost
Projects
NLP
2025
ECommerceFeedbackAI — Customer Feedback Mining & Insights
Ingested & analysed 23K+ e-commerce reviews, reducing manual feedback triage by 80% through automated sentiment analysis, topic modelling, and drift monitoring.
Problem
Unstructured reviews slowed insights; hidden pain-points and fit issues driving returns.
Solution
Built NLP pipelines (clean → sentiment → topic); visualised trends & co-occurrence; drift detection with monitoring KPIs.
Impact
−80% manual analysis, 7 customer themes surfaced, faster product decisions (+15–20%), ~10–12% reduction in returns, ~8–10% retention uplift.
Forecasting
2025
SupplyChainAI — End-to-End AI Supply Chain Platform
Unified demand forecasting, inventory optimisation, and risk intelligence into an interactive Streamlit platform; reduced stockouts 15–30% and improved forecast reliability via P10/P50/P90 fan charts.
Problem
Stockouts & excess holding costs from unreliable forecasts and limited scenario planning.
Solution
Implemented diverse forecasting models; Monte Carlo (200–1000) for (s, S) policies; schema-validated Walmart data; event-driven S&OP risk KPIs.
Impact
−15–30% stockouts, 99%+ data integrity, faster executive decisions via auto-generated summaries, improved developer reproducibility with CLI tools & modular repo.
RecSys
2023
Product Recommendation (Capstone)
Collaborative filtering in Python/Flask; Top-5 accuracy 87%; lightweight API for serving.
Problem
Low attach rate; poor personalization.
Solution
Neighborhood CF + implicit feedback; simple service for A/B tests.
Impact
+CTR on related items; faster discovery of long-tail products.
GenAI
2025
Domain-Aware RAG for Support
Hybrid retrieval + cross-encoder re-rank; guardrails; FRT −41%. Faithfulness & grounding eval.
Problem
Slow, inconsistent answers across support docs.
Solution
BM25+dense (RRF), cross-encoder; PII redaction & refusal policies.
Impact
−41% first response time; higher CSAT on complex queries.
Experience
Data Science Intern — Hightech Masterminds Pty Ltd
Sydney • Feb 2025 – Apr 2025Problem
Churn rising; reporting slow & error-prone due to inconsistent data quality across sources.
Solution
- Designed data validation (nulls, ranges, referential integrity) in ETL.
- Built churn model (LogReg, class-weighted, K-fold CV) with precision/recall monitoring.
- Delivered Tableau dashboards + risk cohort alerts (daily cadence).
Impact
−40%manual cleaning
~85%model accuracy
Dailyreporting (from weekly)
+5–7%projected revenue uplift
Stack
PythonPandasNumPy
scikit-learnSQLTableau
AWS (S3/Redshift)DockerGitHub Actions
Education
Torrens University Australia
Bachelor of Software Engineering (AI)
- Capstone — Product Recommendation: Top-5 accuracy 87% (Python/Flask/Pandas).
- Coursework: ML • DL • NLP • CV • Big Data • Cloud • AI Ethics • Statistics & probability • Data Structure & Algorithms.
Professional Year (IT)
QIBA — Sydney • Jan 2024 – Dec 2024
Australian workplace-ready program focused on ICT practice, communication and industry placement.
- Professional communication & stakeholder updates
- Australian workplace culture & policies
- Agile teamwork & delivery discipline
- Job-readiness & interview prep
- Industry placement experience
- Ethics & compliance in ICT
Certifications
Advanced Data Analytics
Analytical thinking and applied ML with Python.
- EDA & Viz
- Regression/Classification
- Evaluation
- Storytelling
Machine Learning Specialization
- Linear/Logistic
- Bias–variance
- Trees & SVM
- Unsupervised
Deep Learning Specialization
- NNs
- Regularization
- CNNs
- Sequence models
Connect
Open to DS/ML roles & collaborations. Sydney • AU work rights.
I am eager to apply my skills in Data Science and Machine Learning to real-world challenges. With a background in Software Engineering (AI) and hands-on experience across NLP, forecasting, recommender systems, and MLOps, I focus on building solutions that are reliable, scalable, and deliver measurable impact.
I am actively seeking opportunities to contribute as a Data Scientist / ML Engineer. If you are hiring or open to collaboration, I’d be glad to connect and discuss how my expertise can add value to your team or project.