About
Bram was born in Jakarta, Indonesia, and moved to the suburbs of Atlanta when he was 7 years old. He now lives in Tucker, GA with his wife and three kids. He’s tactical, pragmatic, and loves to learn.
What people say about Bram:
His wife says he’s a loving husband, his kids say he’s a fun dad! His parents would say he’s a respectful son, his siblings would say he’s a risk taker, his friends would say he’s encouraging and loyal, his ex-employers/peers would say he’s diligent and gets the job done, and he would say he’s a dreamer and has an optimistic outlook on life.
What does Bram like to do?
- Read bedtime stories to his kids
- Visit the local beer garden with the Wife
- Spend time with family and friends
- Having friends over, smoking meat while drinking beer, and starting a conversation with… remember the time when…
- Golfing, then contemplating on why he still plays because he’s not very good at it
Virtues to live by:
- Empathy - Golden Rule “Treat others the way you want to be treated”
- Decisive - Perform without fail what you set out to do
- Peaceful - Don’t be disturbed at unimportant things, or at random accidents
“He is comfortable presenting new ideas to a large team and is also comfortable and excels at client relations. Bram doesn’t shy away from challenges and stops for feedback along the way. He is extremely effective at making processes – in several areas of a company – more efficient. He stays ahead of the curve in technology, and often implemented processes in this arena that made the day to day operations smoother and time-efficient.”
Chris Wenner - Managing Director, Wennco LLC
Dynamic content generation
A tool that generates custom content based on consumer input
- Utilized LLMs and vector database to generate domain-specific content (RAG method).
- Created pipelines to implement for production use.
Sentence-classifier/style-detection
A tool to detect sentence type and style of writing
- Used ensemble learning to classify sentence type e.g., introduction, call-to-action, etc.
- Used LLMs/Tranformers to identify tone/style of writing, e.g., casual/formal, and elaborate/concise.
Content manager and generation
A tool to create, read, update, and delete content
- Used Streamlit as the UI including authentication
- Created efficiency, usability, and accessibility of content management (before using Google Sheets)
- Generated email suggestions based on phrases and words that resonate with a person’s personality
Personality prediction
A tool to predict Disc type
- Ngram analysis, used to identify correlations between words and personality type
Portfolio Liquidation Forecasting (pre-purchase)
A tool for Debt Buyers to predict portfolio liquidation prior to purchase of using Machine Learning Models
- Ensemble learning which decreased error rate significantly in overall purchase and per portfolio purchase
- Outperformed existing internal models and third-party vendors in which we spend $200k+ annually
Propensity-score
A tool for Collection Agencies to identify and rank-order consumers with the highest likelihood of payment
- Performed resampling strategies dealing with Imbalanced data, created ML-generated features
- Used ensemble learning to increase model performance
Efficient Revenue Recovery - Github
A tool for Collection Agencies to segment and rank accounts by likelihood of payment using Machine Learning Models
- Merge multiple datasets using Pandas, cleaned and explored data, 12+million records and 46+ columns
- Utilized scikit-learn dimensionality reduction: PCA, t-SNE and Truncated SVD in order to cluster classes
- Exhausted multiple classification models, with SVM as the best accuracy
- Created a scoring system using feature importance and manipulated data to segment and rank accounts
- Communicated findings to executives, highlighting areas of potential improvement and ideas for future work
Sarcasm Detection Using News Headlines - Github
Text classification using NLPs to detect sarcasm using news headlines
- Used word embeddings to retain semantic relationships within words
- Utilized recurrent neural network (RNN) with a pre-trained embedding layer GloVe resulting 81% accuracy
- Defined plotting function to visualize model performance
Statistical Testing - Github
Testing if discounts have a statistically significant effect on the number of products customers order
- Checked for data normality visually using Q-Q plot and statistically using SciPy’s normal test
- Performed Welch’s T-test and examined effect size using Cohen’s d
- Summarized results with p-value, confidence level, and the significance of effect size
Languages
Experience
Nashville, TN - Remote (Software) - Product, Data Scientist
Aug 2022 - Present
- Implemented and established data pipelines and data department within organization
- Used ML techniques to optimize personality prediction engine
- Use text mining/analysis, NLP techniques to generate text suggestions (feature) based on DISC types
- Created and implemented sentence category/type detection within email using LLMs
- Generated dynamic content using RAG method - Used LLM api along with vector database api to create custom content
- UI for content management using Streamlit - create, read, update, and delete company content (used gsheet before)
Atlanta, GA (Healthcare Asset Management/Receivables) - Data Scientist II
Nov 2019 - July 2022
- Implemented and established data science method (OSEMN) and department within organization
- Enriched data using NLP/NLTK to predict missing gender, sourced behavioral/demographic data to create consumer segments
- Data enrichment, sourced behavioral data to better understand our consumers (MPI, Esri and ArcGIS toolkit)
- Created Forecasting models to predict portfolio liquidation rate which increased accuracy of liquidation forecast by 100%
Atlanta, GA - Partner, Head of Operations and Trader
Jan 2018 - Sep 2018
- Responsible for trading operations, portfolio management and trade execution
- Assisted portfolio construction on equities and equity options
- Managed and communicated with institutional partner relationships
- Collaborated with CIO on research, trade ideas, products and market letter construction
- Organized internal data and created systems
Atlanta, GA - Head of Technical Analysis and Trader
Feb 2017 - Jan 2018
- Responsible for portfolio construction (equities and options), and market letter construction
- Generated equity / index options for covered call and custom collars for portfolios and clients
- Traded Institutional SMA, and in house SMA strategies, and managed in house Index options Strategy
- Communicated critical data/information to department heads to ensure best practice and outcome
- Presented weekly/monthly reports to investment committee
Atlanta, GA - Equity Research Analyst and Junior Trader
Oct 2016 - Feb 2017
- Researched equities with a base of Graham and Dodd Security Analysis
- Equity screening using sell / buy side research and other publicly available information (regulatory filings)
- Assisted in portfolio construction (equities), data screening and time sensitive research material
Self-employed
Atlanta, GA
Oct 2015 - Oct 2016
- Managed personal portfolio, income generating strategy, using equity options.
Atlanta, GA - North American Operations and Sales Manager
Aug 2009 - Oct 2015
- Generated 60% of sales, breaking $1MM of revenue for North America
- Managed North American operations and communicated with clients to ensure prompt deliveries and satisfaction
- Reported to executives, and collaborated with plant managers to achieve timely deliverables
- Created systems to improve information flow, planning and logistics management
Education/Certification
- Flatiron School: Data Science Immersive
- Statistical analysis, data acquisition, data modeling, machine learning, deep learning, NLP
- Python, SQL, SQLAlchemy, scikit-learn, SciPy, NumPy, Pandas, Keras, Matplotlib, Seaborn
- SQL: DataCamp
- R: DataCamp
- FINRA Series 7, Series 9