Mukund Singh
IT Specialist
4+ projects
SQL
Python
Data Wrangling & Cleaning: 4/10
Predictive Modeling: 4/10
Active 5 hours ago
Invite to interview
Message
Download CVCV
Overview
Technical skills
Roles
Overview
A data engineer-level practitioner (middle) who builds end-to-end data aggregation and ML-ready pipelines with reliable use of PySpark and standard ML tooling. Their strongest proven skill is big-data aggregation and regional/category summarization implemented with Spark window functions and grouped aggregations. Public artifacts do not show robust productionization features such as CI, dependency pinning, data versioning, automated scheduling or comprehensive model tuning and testing.
Technical skills
SQL• Junior • 3 projects
Python• Junior • 4 projects
AI/ML
Spark
Pandas
Scikit-learn
Jupyter Notebook
Analytics
Matplotlib
ETL/ELT
Tableau• 4 projects
Power BI• 2 projects
Middle Data Scientist
Confidence: Medium Data Engineer
A data engineer-level practitioner (middle) who builds end-to-end data aggregation and ML-ready pipelines with reliable use of PySpark and standard ML tooling. Their strongest proven skill is big-data aggregation and regional/category summarization implemented with Spark window functions and grouped aggregations. Public artifacts do not show robust productionization features such as CI, dependency pinning, data versioning, automated scheduling or comprehensive model tuning and testing.
Statistical Rigor
3/10
Correct use of statistics
Basic predictive-evaluation practices are present (precision/recall/F1, stratified split, class-weighting) but formal statistical rigor, uncertainty quantification, hypothesis testing or multiple-comparison controls are missing.
Evidence
bank-marketing-prediction/notebook.ipynb: use of classification_report and f1_score; stratified train_test_split
bank-marketing-prediction/notebook.ipynb: class_weight='balanced' in model constructors
Data Wrangling & Cleaning
4/10
Preparing and cleaning data
Reasonable data ingestion and cleaning steps are implemented including CSV ingestion with schema inference in Spark, null checks and label encoding; however there is limited advanced provenance, outlier handling or leakage-prevention beyond stratified splitting.
Evidence
pyspark-sales-analysis/sales_analysis.py: spark.read.csv(..., inferSchema=True) and groupBy aggregations
bank-marketing-prediction/notebook.ipynb: df.isnull().sum() and LabelEncoder usage for categorical columns
Exploratory Analysis & Visualization
3/10
Exploring and visualizing data
Exploratory outputs and visualizations exist (head, value counts, feature importance bar chart) but written interpretation is minimal and EDA is not structured around clear hypotheses.
Evidence
bank-marketing-prediction/notebook.ipynb: plt barh of feature importances and printed head/value_counts
pyspark-sales-analysis/sales_analysis.py: .show() outputs for summaries and top categories
Predictive Modeling
4/10
Building models that predict
Predictive workflow follows baseline-first discipline using Logistic Regression and Random Forest with class balancing and stratified splits; missing are systematic hyperparameter tuning, cross-validation schemes beyond a single split, calibration or detailed error analysis.
Evidence
bank-marketing-prediction/notebook.ipynb: LogisticRegression and RandomForestClassifier training with random_state and class_weight
bank-marketing-prediction/notebook.ipynb: classification_report and confusion matrix usage
Business Insight & Impact
2/10
Turning analysis into business value
There are surface-level links to business signals via feature importances (call duration importance is surfaced) but there is little concrete cost/impact analysis, decision-threshold reasoning or action-oriented recommendations.
Evidence
bank-marketing-prediction/notebook.ipynb: printed feature importances showing duration as top predictor
Reproducibility & Notebook Hygiene
3/10
Clean, repeatable analysis
Some reproducibility measures exist (random_state seeds, standalone PySpark script), but pinned environments, dependency manifests, data versioning, automated pipelines or testing are absent.
Evidence
bank-marketing-prediction/notebook.ipynb: train_test_split(random_state=42) and RandomForestClassifier(random_state=42)
pyspark-sales-analysis/sales_analysis.py: SparkSession.builder and direct CSV read/write for pipeline-style execution
Expertise
Big Data• Middle
Industries
Commerce• Middle
Financial Services• Middle
Technologies
Spark
Jupyter Notebook
Scikit-learn
Matplotlib
Pandas
ETL/ELT
Python• mentioned only
Recommendations
- Develop production ETL pipelines that include dependency manifests (requirements.txt or environment.yml), a simple CI workflow, and parameterized run scripts for Spark jobs.
- Add reproducibility and data provenance: pin package versions, include a small data sample or data catalog entries, and track raw-to-processed transformations.
- Introduce systematic model validation: cross-validation, hyperparameter search (Optuna/Hyperopt), threshold tuning with business-cost analysis and calibration checks.
- Convert iterative notebooks into modular scripts or lightweight packages and add unit tests for core data transformation functions.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Junior AI/ML Engineer
Confidence: Medium Data-centric
Data-centric engineer (Middle) focused on PySpark sales and retail analytics. The strongest proven skill is building PySpark data pipelines and windowed aggregations for region- and category-level sales analysis as implemented in pyspark-sales-analysis/sales_analysis.py. There is no public evidence of model training, experiment tracking, production serving, automated tests or robust error handling.
Model Architecture & Training
How well models are designed and trained
Not evidenced in public code
Data Pipeline & Feature Engineering
4/10
How data is prepared for models
Data ingestion, aggregation and windowed ranking implemented with PySpark DataFrame API and Window functions for sales analysis.
Evidence
mukund2singh-dev/pyspark-sales-analysis/sales_analysis.py: spark.read.csv('sales_data.csv', header=True, inferSchema=True)
mukund2singh-dev/pyspark-sales-analysis/sales_analysis.py: df.groupBy('Region').agg(_sum('Sales').alias('Total_Sales'), _sum('Profit').alias('Total_Profit'), count('OrderID').alias('Order_Count'))
mukund2singh-dev/pyspark-sales-analysis/sales_analysis.py: Window.partitionBy('Region').orderBy(desc('Category_Sales')) and rank().over(window_spec) to produce top categories per region
Experimentation & Evaluation
1/10
How results are measured and tested
Basic interactive inspection and printed outputs for EDA without formal experiment tracking, validation splits or metrics frameworks.
MLOps & Deployment
1/10
How models are shipped to production
Minimal operational work: writes CSV outputs and manages Spark session lifecycle but no serving, versioning, CI/CD or monitoring artifacts.
Computational Efficiency
1/10
How efficiently computing resources are used
Uses Spark (distributed compute) and windowing but no explicit performance tuning, caching, partitioning, profiling or GPU/quantization work.
Research Depth & Innovation
Depth of research and new ideas
Not evidenced in public code
Expertise
E-commerce & Retail AI• Junior
Finance & FinTech AI• Junior
Industries
Commerce• Middle
Financial Services• Junior
Recommendations
- Build end-to-end PySpark ETL pipelines that include parameterization, input validation and configurable output paths.
- Develop repeatable analytics jobs with unit tests, logging and simple CI to harden workflows for production use.
- Extend skills toward deployment and observability by adding lightweight job scheduling and monitoring (Airflow/Argo/Prometheus) and experiment tracking if moving to modeling work.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories:
Junior Backend Developer
Confidence: Medium Data Platform
Junior backend developer focused on data processing and simple CRUD tooling with a practical strength in writing small-scale data jobs. Proven skill in PySpark data aggregation and windowed ranking as implemented in sales_analysis.py. No evidence of production-grade API design, database migration history, automated tests, or robust security and observability practices.
API Design
How well APIs are designed
Not evidenced in public code
Data Layer & Database
2/10
Working with databases
Basic data-layer work is present via CSV ingestion and simple file-based persistence but there is no migration history, transactional control, or DB isolation awareness.
Scalability & Performance
2/10
Handling load and speed
Some awareness of scalable data processing is visible through use of Spark and window functions, but there are no explicit performance tuning, caching, connection pooling or load-handling patterns.
System Architecture
1/10
Overall system structure
Design is single-file, script and console-focused with minimal service decomposition or configuration/secret management.
Security & Auth
1/10
Protecting data and access
Very limited security posture; authentication exists but is hardcoded and there is no evidence of secure input handling, token lifecycle, or secrets management.
Reliability & Observability
2/10
Stability and monitoring
Minimal reliability hygiene is present such as try-with-resources for file IO and explicit spark.stop(), but there are no structured logs, retries/backoff, timeouts, or observability hooks.
Expertise
Databases & Vector Storage• Junior
Industries
Commerce• Middle
Education• Junior
Recommendations
- Develop small to medium PySpark ETL and analytics jobs that run on managed Spark clusters or scheduled pipelines.
- Build and migrate simple CRUD prototypes to a relational database with proper migrations and transactions to replace file-based storage.
- Work on adding basic observability and security practices such as structured logging, unit tests, input validation, and secret management
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories: