- Break down a business problem into smaller components and implement ML/NLP approaches to achieve the maximum accuracy.
- Work with others to select the appropriate model for the data provided, resolve various shortcomings of the models such as over fitting, high error rate, poor F1 scores etc.
- Read literature and come with out of the box solutions for hard problems.
Desired skills and experience :
- Strong NLP skills which include Language Modelling, POS tagging, PCFG, Named Entity Recognition, Co reference Resolution, Question Answering.
- Hands on and good understanding of various classification techniques such as Clustering, Logistic Regression, CRFs, MEMM, Neural Networks, SVMs, Decision Trees.
- Hands on experience of various IR methods including Data Pre processing, Similarity Measures, Tf-Idf, Ranked Retrieval, Indexing.
- Knowledge of linear algebra and dimensional reduction algorithms such as PCA, Locality Sensitive Hashing, Auto encoders.
- Programming experience in Java or Python or both is a must.
- Strong interpersonal and communication skills: ability to tell a clear, concise, actionable story to the folks across various levels of the company.
Nice to have skills :
- Knowledge of optimization techniques like L-BFGS, Gradient Descend, NM method, Back propagation etc and when to use which one.
- Experience of working with large amount of data using any of the following Spark, Storm, Hadoop, Mlib, Mahout.
- Knowledge of Deep Learning techniques Autoencoders, Stacked RBMs, RNNs, CNNs etc and libraries like Theano, Torch, CuDNN, deeplearning4j, Caffe etc
- Working experience of R, Matlab or Octave.